1 karl 1.1.2.1 Using the CIM/XML Pull Operations
2
3 STATUS
4
|
5 karl 1.1.2.4 <<< The TODO section is being maintained during the review and checkin process
|
6 karl 1.1.2.1 to keep track of problems, errors, notes, etc. Must be deleted before
7 checkin to head of tree. Please feel free to add notes, etc in this
8 section as you review/test.>>>>>>
9
|
10 karl 1.1.2.4 TODO list:
|
11 karl 1.1.2.3 1. Binary operation from OOP. Need to add counter to binary
12 protocol to be able to count objects in response. Generates
13 warnings in things like messageserializer and does not work with
|
14 karl 1.1.2.31 OOP right now. Fixed by converting to XML. Concluded that we do not
15 need to do this. The binary response is not really used often
16 in the current environment So double mapping it is not a major issue.
17 Leave this as FUTURE
|
18 karl 1.1.2.23 2. OpenExecQuery - Code is incomplete in that it does not include the
|
19 karl 1.1.2.31 return from the exec query function to the aggregator yet. Code for Pull
20 part of OpenQueryInstancesRequest a) should be part of
|
21 karl 1.1.2.4 the common CIMOperationRequestDispatcher execCommon code.
|
22 karl 1.1.2.31 3. The changes to WQLCIMOperationRequestDispatcher and CQL... for handling
|
23 karl 1.1.2.4 pull not completed so we feed the responses back to the EnmerationContext
24 queues
|
25 karl 1.1.2.31 4. Minor TODOs, diagnostics, etc. still in the code
26 5. Extension to avoid double move of objects in CIMResponseData (one
|
27 karl 1.1.2.3 into enumerationContext queue and second to new cimResponseData for
28 response. Want to avoid second move by extending Open/Pull response
29 messages to include count and CIMResponse data to count objects out
|
30 karl 1.1.2.23 of queue when converting (avoids the second move). This would mean
31 extending the output writers to allow the count field to be supplied
|
32 karl 1.1.2.30 so they would only create output for up to the count supplied.(Make
33 this future beyond bug 9676)
|
34 karl 1.1.2.31 6. Add more static tests (currently only OpenEnumerateInstances and
|
35 karl 1.1.2.30 OpenEnumerateInstanceNames covered).
|
36 karl 1.1.2.26 Low priority because most of this except exact XML covered in other
|
37 karl 1.1.2.28 tests. note that we can only add very limited tests since a) no
|
38 karl 1.1.2.31 way to test multiple operations in sequence(i.e. open/pull) and
|
39 karl 1.1.2.28 cannot be sure how many responses will come back in all cases even
40 for open.
|
41 karl 1.1.2.31 7. Could clean up the CIMOperationRequestDecoder.cpp rejectCIMException
42 further since there is some overlap. VERY Low priority.
43 8. Make enumContextTable hashtable dynamic and create entirely within
|
44 karl 1.1.2.30 CIMOperationRequestDispatcher.cpp. Today it has a static size
|
45 karl 1.1.2.31 9. Correct issue between operations and HTTP where we are sending
46 trailers with exceptions. Modify response mechanisms so that we
47 set non-chunked for all responses where we send error responses to
48 avoid the trailers.
49 10. It would be more efficient in dispatcher to always use exception for
50 rejects and change the _reject functions so that they never return
51 when they reject. This is VERY LOW PRIORITY and primarily saves
52 a few lines of code in the reject functions and their calls. Means we
53 would code.
54 _rejectIfEnumerationToBroad(...)
55 _rejectIfThisParameterIncorrect(...)
56
57 instead of
58 if (_rejectIfEnum...)
59 {
60 return true
61 }
62 It would mean that the method trace for the handlers would not return
63 an exit if we rejected. LOW PRIORITY
64 11. We have reduced the number of template function greatly. However there
65 are still several around the task of distributing requests to the
66 karl 1.1.2.31 multiple providers. In fact there are two similiar but different for
67 the associators(i.e. assoc, ref, etc.) functions and the parallel
68 openAssoc functions. It would be nice to consolidate that logic and
69 further to try to create a non-template form for those functions. LOW
70 PRIORITY.
71 13. External runtime variables. Proposing that they be fixed for this release
72 rather than set by configuration. Making this a separate bug.
73 See bug 9819 for the changes to cover this.
|
74 karl 1.1.2.26
75 NEXT TASKS:
|
76 karl 1.1.2.31 a. finish 8, 9 3 above
77
78 9 May 2104
79 1. Converted several template functions in CIMOperationRequestDispatcher
80 to real functions
81 2. Added intermediate CIMMessage for CIMPullOperationRequestMessage since
82 all of the pulls have exactly the same data. This allowed us to
83 convert the template function processPullRequest from template to
84 real function
85 3. Added intermediate CIMMessage for CIMOpenOperationRequestMessage betweeh\n
86 all of the Opens and the CIMRequestData message. This will allow
87 common code to test all of these variables in all 6
88 CIMOperationRequestDispatcher handleOpen... functions
89 3. Added variable in CIMOperationRequestDispatcher handleEnqueue to make
90 releasing the original request conditional on return from the handle...
91 for open... and pull... handlers.
92 This allowed us to drop one place where we were creating a new request msg
93 for the delayed response and use the original request. This meant that
94 we could convert the issueOpenOrPullResponse from template function to
95 real function.
96 4. Moved open and pull constructors in CIMMessage.h to CIMMessage.cpp. NOTE:
97 karl 1.1.2.31 openQueryInstances not moved. Will do when we get this funciton running.
98 5. Added test for correct namespace in pull operations. Note that it is
99 never really used but since supplied with request, we test to be sure it is
100 same as open request namespace. added internationalized exception for this.
101 6. We should probably add same test to close just as an input parameter
102 validation.
103 6. Modified processPullRequest to close the enumeration context if
104 incorrect response type received. Before we just returned exception and
105 left context open for next request. Closing is much more logical since the
106 client really did send an erronous request.
107 7. Modified PullErrors to correctly test above changed in behavior.
108 8. Removed a significant number of diagnostic messages.
109
|
110 karl 1.1.2.23
|
111 karl 1.1.2.30 4 May 2014
112 1. Clean interface with EnumerationContext timeout thread
113 2. Remove more diagnostics
114 3. Correct issue with enumerationContext Release that was causing
115 failures during nightly tests.
116
|
117 karl 1.1.2.29 2 May 2014
118 1. Remove Diagnostics from ResponseData.cpp
119 2. Correct setup and start of EnumerationContextTimerThread
120 3. Change limit on max number of enum contexts to be set from Dispatcher.
121 4. Increase size of hashtable for enum contexts.
122 5. Corrected issue causing failures in removeContext. Now runs through nightly
123 tests most of the time.
|
124 karl 1.1.2.28 1 May 2014
125 1. Cleanup of code and variable names.
126 2. Removed namespace variable from EnumerationContext. It was never used.
127 3. Added test for max number of open enumerations. Right now it is fixed
128 and set about 100. Note that this is tied to the size of hash table
129 so that there is a issue with more dynamic setting of max and the
130 hash table size.
131 4. Removed several unused functions.
132 5. Found one possible issue causing thread conflict but not the core one
133 yet.
134
135 30 April
136 1. Added exception response to dispatcher for Limits exceeded on
137 CreateEnumerationContext. Returns Exceeded server limits CIMException.
138 2. Fixed possible deadlock where we were not unlocking EnumerationContext
139 under error condition
140
141
|
142 karl 1.1.2.27 27 April - mergeout to head of tree
143 1. Mergeout of the code to the head of tree. Documented that mergeout in
144 the bug
145
146 27 April 2014
147 1. Clean up in a number of areas, mostly just code without changing
148 functionality.
149 2. Modified CIMMessage.h slightly to clarify names of the pull intermediate
150 messages (the common open and pull response);
151 3. Removed a number of diagnostics from CIMResponseData
152 4. Added new function to ResponseStressTestProvider to be able to terminate
153 output after a defined number of objects with a CIMException.
154 5. Removed a number of other TODOs
155 6. Corrected a couple of errors in the server message bundle.
156 7. Modified ResponseEncoder to only output the error for pull errors.
157 It was errounously outputting the EndOfSequence, EnumerationContext and
158 the error. This way, only the error is sent which is spec.
159 8. Removed number of TODOs and cleaned up a couple of others.
160 9. Added static tests for OpenEnumerateInstanceNames.
161 10. Modified a number of the temporary diagnostics to be permanent by
162 changing them to PEGASUS_DEBUG_ASSERTs
163 karl 1.1.2.27
|
164 karl 1.1.2.24 22 April 2014
165 1. Clean up some messages.
166 2. Still trying to find issue that randomly keeps messages in timeout
167 queue long after they have been closed and returned.
|
168 karl 1.1.2.25 3. Fixed memory loss error.
|
169 karl 1.1.2.26 4. Added zero reject for OperationTimeout because that effectively disables
170 the whole timeout mechanism so that enumerations could be started, never
171 completed by the client and would just hang around in memory. That
172 would be a significant DDOS attack (start very large enum with no
173 timeout) and never do pulls after open.
|
174 karl 1.1.2.25
|
175 karl 1.1.2.23 21 April 2014
|
176 karl 1.1.2.27 1. Reorganized code in CIMOperationRequestDecoder.cpp ...Iparam classes.
|
177 karl 1.1.2.23 Cleaned up the rejectCIMException code.
178 2. Cleanup some issues in the EnumerationContextTimeout Thread.
179 3. Fixed issue in pullop where it was using zero as default for default
180 operation timeout where it should have been NULL, the DMTF specified
181 default.
182 4. Enabled the test to reject 0 as operation timeout on a request and but
183 controled by a #define in CIMOperationRequestDispatcher.cpp
184 5. Added test for excessive new OperationContext entries in table.
185
186 18 April 2014
187 1. Internationalize remaining messages in CIMOperationRequestDecoder.cpp and
188 clean up the decoders for some of the messages so they use the
189 ...Iparam classes. Note that we did not clean up all of the old message
190 decoders, just the new ones and we did not clean up the Count message
191 since it is deprecated. TODO clean these up so the decoder is really
192 consistent.
193 2. Enabled the EnumerationContext Timer and removed code for alternate
194 non-thread solution.
|
195 karl 1.1.2.16
|
196 karl 1.1.2.21 17 April 2014
197 1. Reorganized the lock/unlock for handling the EnumerationContext because
198 there were thread windows (i.e. using the context after it was released
199 by other thread).
200 2. Modified code for the InteroperationTimerThread and deleted old code that
201 allowed this whole thread to by bypassed.
202 3. Internationalized a number of CIMOperationRequestDispatcher CIMException
203 response messages.
|
204 karl 1.1.2.22 4. Fixed stress tests in Client/tests/pullop that were getting timeout. The
205 issue was not the pull code but that fact that the idle client timeout
206 appears to be measureing to the end of the operation and the operation
207 was taking more than 40 seconds which is the default timeout. The idle
208 timeout takes no account of the fact that chunks are being regularly
209 delivered.
210 5. Several minor code changes.
211 NOTE: At this point the code passes extensive tests of the pull operations
212 both in OOP and none OOP mode (but with lots of diagnostics that may
213 be affecting timing).
|
214 karl 1.1.2.21
|
215 karl 1.1.2.20 10 April 2014
216 1. Fix issues causing failure with repeated pull operations.
217 2. Consolidate some of the CIMOperationRequestDispatcher.cpp template code
218
|
219 karl 1.1.2.19 7 April 2104
220 MergeOut of head of tree.
221
222 7 April 2014
223 1. Permanently fixed issue with getting correct number of objects on response.
224 This was introduced when did a temp fix to the OOP processing, in particular
225 removed the condition variable wait in the CIMOperationDispatcher handle
226 functions for opens and pulls. This moved the functionality to kick off
227 open and pull responses to the provider threads if there are not enough
228 objects when the handle is processed. Also added a variable to allow us to
229 test with either a) responses are required to satisfy the original
230 request size or) b, responses are required only to return some objects.
231 We will add statistics to see which of these works best.
232 2. Modified WsmProcessor EnumerationContext class since the name conflicts
233 with the pull Operation EnumerationContext class.
234 3. Created a new intermediate level of CIMMessage, the CIMPullResponseData
235 Message so that all of the open and Pull responses can use common code
236 since there are only two variables (endOfSequence and EnumerationContext)
237 and they are common across all the open and pull responses.
238 4. Removed a number of diagnostics.
239 5. Cleaned up the Dispcatcher so that the open and pulls have common response
240 karl 1.1.2.19 code and that code can be used from the dispatcher handle functions and
241 the provider response functions.
242
|
243 karl 1.1.2.18 31 March 2014 - Checkin
244 1. Fixed issues in OOP processing of pull operations, in particular
245 issues with cimxml output format when processed through the
246 *InternalXmlEncoder functions.
247 2. Clean up some of the internalXml functionality
248 3. Found issues causing timeout with a particular provider. The issue
249 is that the dispatcher and monitor end up using the same thread so the
250 condition variable in the dispatcher thread stops the monitor. Turned
251 off the conditionVariable in getCache for the moment which means that
252 we get number of responses for open... with 0 objects before the
253 providers can begin to respond. This is only for test.
254 4. Added some statistics for enumerations and display the statistics
255 when we close the server (same as cache statistics)
256
257 12 March 2014 - Mergeout and Mergein
258 1. Mergeout to head of tree for this date and mergein for patch update
259 to bug 9676
260 2. Extensions to pullop tests program and tests.
261 3. Added some diagnostics in looking for OOP issue.
262 4. Removed a number of diagnostics messages and cleaned up code in
263 dispatcher to simplify pull operation processing.
264 karl 1.1.2.18
|
265 karl 1.1.2.17 15 December 2013
266 1. Mergeout and mergein up to 15 December 2013
267 2. Clean up issues from tests documented in bug 9676 last week.
268 3. Clean up some code in dispatcher
269 4. Remove the filter function from ResponseStressc++Provider.
270
|
271 karl 1.1.2.16 21 November 2013
272 1. Mergeout from head of tree to 21 November 2013.
|
273 karl 1.1.2.15
274 18 November 2013
275 1. Cleanup of a bunch of minor errors and completion of all of the code for
276 the openQueryInstances except for the PullInstances in Dispatcher and
277 the aggregator function.
278 2. OpenqueryInstances added to cimcli.
|
279 karl 1.1.2.14
280 13 October 2013 CVS branch update.
281 1. Integrated bug 9786 into the branch. Note that we need to test the
282 generated statistics.
283 2. Mergeout executed to update to head of tree as of 8:00 am 13 October 2013.
284 3. Cleaned up several errors in OOP processing. Note that there is at least
285 one issue left when we to a pull on ManagedElement in at least one of the
286 namespaces.
287 4. Cleaned up some of the outstanding diagnostic code
288 5. Generally passes all tests except for one test of pullop where it is trying
289 to pull enum instances CIM_ManagedElement from a particular namespace.
|
290 karl 1.1.2.12
|
291 karl 1.1.2.16 NOTE: I did not make comments here for changes in October despite the fact
292 that I did 2 mergouts, number of fixes, and a mergein.
293
|
294 karl 1.1.2.12 30 September 2013 - CVS Update
295 Mergeout head of tree up to 29 September 2013.
|
296 karl 1.1.2.10
297 29 September 2013. CVS update.
298 1. Modified calls to statisticalData.cpp to a) directly call with request
299 type, b) incorporate the open, pull, etc. messages. However, since these
300 are not part of the CIM class, we must do something special with them.
|
301 karl 1.1.2.13 See bug 9785 for full solution to this issue.
|
302 karl 1.1.2.10 2. Corrected OOP interface to enable new flag to indicate internal operations
303 and set host, etc.
304 3. Add code to CQLOperationsDispatcher and WQLOperationDispatcher to clean
305 up CIMResponseDataCounter after filtering.
306 4. Modified ProviderAgent to set Host info for some pull operations.
307 5. Added new flag to CIMBinMsgSerializer and Deserializer.
|
308 karl 1.1.2.8
|
309 karl 1.1.2.9 17 September 2013 CVS update (Actually two different updates over 3 days)
310 1. Clean up some issues in CIMMessage.h and CIMMessage.cpp
311 2. Extend OpenExecQuery to WQL and CQL processors but return not complete
312 3. Remove memory leak in EnumerationContext and EnumerationContextTable
313 handling.
314 4. Created template functions for much of the pull operations.
315 5. Reversed order of queryLanguage and query (and changed names to match
316 execQuery) in client and server. Note that these are the execQuery
317 WQL and CQL filters and NOT FQL filters.
318 6. Some code cleanup in dispatcher
319 7. Today, passes all tests in pullop but issue in alltests. For some reason
320 not finding CIMObjectManager instance. Also, leaves enumeration contexts
321 if client terminates since cleanup thread not operating.
|
322 karl 1.1.2.11 8. XML from OOP not correctly processed.
|
323 karl 1.1.2.9
|
324 karl 1.1.2.8 14 September 2013 CVS update
325 Merged out up to 25 August. Cleaned up all operations and standardized code.
326 At this point the non pull operations code is in a set of templates but the
327 pull is not yet.
328 Fixed a significant number of problems so that it appears that the operations
329 except for OpenExecQuery run stably, at least with the pullop test program.
330 Note that there is a problem in that the Interop control provider is not
331 returning its singleton wbemserver object for some reason. Causes a test
332 failure
|
333 karl 1.1.2.7
334 Fixed for 16 June CVS Update
335 1. Cleaned up the enumerationContext and Table release functions and tested
336 to confirm that we do not lose memory in either normal sequences or
337 sequences that close early. Cleaned up pullop and added more tests
|
338 karl 1.1.2.8 Taged Before: PREAUG25UPDATE and after POSTAUG25UPDATE
|
339 karl 1.1.2.4
340 Fixed for 9 June CVS update
341 1. Cleaned up code for OpenQueryInstances. Note that this is incomplete.
342 No support in WQL or CQL Operations
343 2.
344
345 What was fixed for 5 June checkin.
|
346 karl 1.1.2.3 1. Extended ResponseTest MOF for for both CMPI and C++ subclasses
347 2. Fixed issues with pullop.
348 3. Fixed temp issue with CIMResponseData size by putting in mutex. That
349 is not a permanent fix but it gets around issue probably in the control
350 of the move logic that meant counts were off.
351 4. Fixed issues in Dispatcher so that associator code works. Still messy
352 code in the dispatcher.
353 5. Changed name of Enumerationtable.h & cpp to EnumerationContextTable.*
354 6 Changed name of ResponseStressTest module, classes, etc.
355
356 TAG: TASK_PEP317_5JUNE_2013_2
357
|
358 karl 1.1.2.2 2 June 2013
|
359 karl 1.1.2.1
360 Issues - KS
361
|
362 karl 1.1.2.15 - Still way to many TODO and KS comments and KS_TEMPS. Removing bit by bit.
|
363 karl 1.1.2.1
|
364 karl 1.1.2.15 - Runtime variable connection for the config parameters not installed. That
365 has been made into a separate bug (see bug 9819)
|
366 karl 1.1.2.1
367 5. Issue with the threaded timer. For some reason during tests it
368 eventually calls the timer thread with trash for the parm (which is
369 pointer to the EnumerationTable object). Caught because we do a valid
370 test at beginning of the function.
371
|
372 karl 1.1.2.2 6. Still using the templates in CIMOperationRequestDispatcher to simplify
373 the handle... processing.
374
375 7. I think I have a way around the double move of objects in the
376 EnumerationContext so that the outputter will just take a defined number
377 of objects directly from the gathering cache and save the second move.
378
|
379 karl 1.1.2.15 8. Not yet passing all tests but getting closer now. The major test that is
380 causing an error today is the execution of a full enumeration with the
381 forceProviders = true. This causes a client timeout sometimes.
|
382 karl 1.1.2.2
383
384
|
385 karl 1.1.2.1 ===========================================
386
387 OVERVIEW:
388
389 The operation extensions for pull operations defined in the DMTF specification
390 DSP0200 V 1.4 were implemented in Pegasus effective Pegasus version 2.11
391 including Client and Server.
392
393 These operations extend the CIM/XML individual operations to operation
394 sequences where the server must maintain state between operations in a
395 sequence and the client must execute multiple operations to get the full
396 set of instances or instance paths.
397
398 The following new CIM/XML operations as defined in DSP0200 are included;
399
400 -OpenEnumerateInstances
401 -openEnumerateInstancePaths
402 -OpenReferenceInstances
403 -OpenReferenceInstancePaths
404 -OpenAssociatiorInstances
405 -OpenAssociatorInstancePaths
|
406 karl 1.1.2.16 -OpenQueryInstances
|
407 karl 1.1.2.1 -PullInstancesWithPath
408 -PullInstancePaths
|
409 karl 1.1.2.16 -PullInstances
|
410 karl 1.1.2.1 -CloseEnumeration
411 -EnumerationCount
|
412 karl 1.1.2.2 OpenExecQuery
|
413 karl 1.1.2.1
414 The following operations have not been implemented in this version of Pegasus:
415
416 -OpenQueryInstances
417
418 The following limitations on the implementation exist;
419
420 1. The filterQueryLanguage and filterQuery parameters are processed by
421 the Pegasus client but the server returns error if there is any data in
|
422 karl 1.1.2.2 either parameter. This work does not include the development of the
423 query language. Note that a separate effort to extend Pegasus to use
424 the DMTF FQL query language is in process.
|
425 karl 1.1.2.1
426 2. The input parameter continueOnError is processed correctly by the client
427 but the Pegasus server only provides for false since the server does not
428 include logic to continue processing responses after an error is
429 encountered.
430 This is consistent with the statement in the specification that use of
431 this functionality is optional and the fact that the DMTF agrees that all
432 of the issues of continuing after errors have not been clarified.
433
434 3. The operation enumerationCount is not processed by the server today since
435 a) really getting the count would be the same cost as the corresponding
436 enumeration, b) the server does not include a history or estimating
437 mechanism for this to date.
|
438 karl 1.1.2.2 NOTE: After a through review as part of the development of the next version
439 of CMPI we have concluded that this operation is probably not worth the
440 effort. Since it is optional, Pegasus will only return the unknown status
441 at this point
|
442 karl 1.1.2.1
443 Since the concept of sequences of operations linked together (open, pull, close)
444 is a major extension to the original CIM/XML operation concept of completely
445 independent operations several new pieces of functionality are implemented
446 to control interOperationTimeouts, counts of objects to be returned, etc.
447
448 TBD - Review this
449
450 CLIENT
451
452 The new operations follow the same pattern as the APIs for existing operations
453 in that:
454
455 1. All errors are handled as CIMException and Exception
456
457 2. The means of inputting parameters are the same except that there are
458 significantly more input parameters with the open operations and for the
459 first time operations return parameters as well as objects in the
460 response. Specifically the open and pull operations return values for
461 enumerationContext which is the identity for a pull sequence and
462 endOfSequence which is the marker the server sends in open and pull
463 karl 1.1.2.1 responses when it has no more objects to send.
464
465 The significant differences include:
466
467 1. Processing of parameters on responses (i.e. the endOfSequence and
468 enumerationContext parameters are returned for open and pull operations).
469
470 2. Numeric arguments (Uint32 and Uint64 include the option of NULL in some
471 cases so they are packaged inside classes Uint32Arg and Uint64Arg in the
472 client api.
473
474 3. The association and reference operations ONLY process instances. They do
475 not include the capability to return classes like reference and associator
476 do and therefore return CIMInstance rather than CIMObject.
477
478 4. Paths are returned in all cases (i.e OpenEnumerateInstances and
479 PullInstancesWithPath where they were not with EnumeratInstances.
480
481 5. The client must maintain state between operations in a sequence (using
482 the enumerationContext parameter).
483
484 karl 1.1.2.1 TBD- Are there more differences.
485
486
487 SERVER
488
489 The Pegasus server attempts to always deliver the requested number of objects
490 for any open or pull request (the specification allows for the server to
491 deliver less than the requested number of objects and specifically to return
492 zero objects on open). We felt that it was worth any extra cost in processing
493 to provide the client with exactly what it had requested.
494
495 The pegasus server always closes an enumeration sequence upon receipt of any
496 error from the providers, repository, etc. Therefore the server will reject
|
497 karl 1.1.2.2 any request that has continueOnError = true;
498
499 Expansion to allow the continue on error may be added in a future version.
500 In any case, the whole purpose of the continue on error is really to allow
501 input from good providers to be mixed with providers that return errors so
502 that generally this would mean simply changing the logic in the return mechanism
503 to not shutdown when an error is received from any given provider.
504
505 Generally we do not believe that the providers need to do much more in the
506 future to support the continueOnError other than possibly allowing the provider
507 to continue processing after it has received an error.
|
508 karl 1.1.2.1
509 PROVIDERS
510
511 This implementation requires NO changes to the existing providers. The
512 provider APIs operate just as they do with the original operations.
513
514 Because the server processing is different however, there may be some
515 behavior differences primarily because the client now controls the speed of
516 delivery of objects.
517
518 In previous versions of Pegasus, the server attempts to deliver objects as
519 rapidly as then can be put on the network. In the case of HTTP chunked requests
520 they are delivered in chunks of about 100 objects. The primary delay for the
521 providers was the processing of each segment through the server. The server
522 is blocked so that no other segment can proceed through the server until that
523 segment is processed and sent on the network.
524 In the case of non-chunkedresponses, they are completely gathered in the serve
525 and then delivered as one non-chunked response. There were no delays for the
526 providers, just lots of possible memory use in the server.
527
528 The responses from providers (delivered through the deliver(...) interface are
529 karl 1.1.2.1 gathered into segments of about 100 objects and this group of objects is moved
530 through the server to be delivered to the client.
531
532 However with the inclusion of the pull operations, The segments of objects
533 from the providers are cached in the server response path until the
534 maxObjectCount for that request (open or pull) and that number returned in a
535 non-chunked response. Thus, if the client is slow to issue pull requests,
536 the providers might be delayed at some point to reduce memory usage in the
537 server (the delay appears as slow response tothe deliver operation).
538
539 In other words, the time to process large sets of responses from the provider
540 now depends on the speed of handling the client.
541
542 It is important to remember in developing providers that the Pegasus server
543 can most efficiently process responses if they are passed from the provider
544 to the server individually or in small arrays of objects rather than the
545 provider gathering very large arrays of objects and sending them to the
546 server.
547
|
548 karl 1.1.2.2 NEXT GENERATION PROVIDERS
549 KS_TODO
550
|
551 karl 1.1.2.1 CONFIGURATION PARAMETERS
552
553 The server includes several configuration parameters to set limits on the
554 processing of pull operations. All of these configuration parameters are
555 compile time parameters rather than runtime.
556
557 1. Maximum value of minimum interoperation time. This parameter defines the
558 maximum time allowed between the return of an open or pull response and
559 the receipt of the next pull or a close operation before the server may
560 close the enumeration. The specification allows the server to set a
561 maximum interoperation time and refuse open requests that with requested
562 operationTimeout greater than that time.
563 CIM_ERR_INVALID_OPERATION_TIMEOUT
564
565 This value is set with the Pegasus environment variable
566 PEGASUS_PULL....
567
568 2. Maximum objects returned in a single open or pull operation. The server
569 can set a maximum limit on the number of objects that can be returned in
570 a single open or pull oepration with the maxObjectCount parameter.
571
572 karl 1.1.2.1 3. Whether the server allows 0 as an interoperation timeout value. The value
573 zero is s special value for the interoperationTimeout in that it tells the
574 server to not timeout any enumeration sequence.
575
576 With this value for interoperationTimeout, the only way to close an
577 enumeration sequence is to complete all of the pulls or issue the close.
578 If for some reason the sequence is not completed, that enumeration context
579 would remain open indefinitly. Since in Pegasus any open enumeration
580 context uses resources (the context object and any provider resposnes that
581 have not yet been issued in a response) it would appear that most
582 platforms would not want to allow the existence of enumeration contexts
583 that cannot be closed by the server.
584
585 4, maximum consecutive pull requests with 0 maxObjectCount. The use of the
586 pull operation with maxObjectCount set to zero could be used to keep an
587 enumeration context open indefinitly (this tells the server to restart the
588 interoperationTimeout but not send any objects in the response). Therefore the
589 specification allows for the server setting maximum limits on this behavior
590 and returning the error CIM_ERR_SERVER_LIMITS_EXCEEDED if this limit is
591 exceeded.
592 Note that this is maximum CONSECUTIVE pulls so that issuing a pull with
593 karl 1.1.2.1 a non-zero count resets this counter.
594
595 KS-TBD - Is this really logical since we can still block by just issuing
596 lots of zero request and an occansional request for one object.
597
598 Pegaus sets the value of this limit to 1000 and allows the implementer to
599 modify it with the PEGASUS_MAXIMUM_ZERO_OBJECTCOUNT environment variable.
600
601 5. Default operationTimeout -
602
603 The default of this parameter is to refuse operat
604
605 In the current release of Pegasus these are all compile time parameters.
|
606 karl 1.1.2.11
607
608 NOTES On working with task branch.
609
610 Merge out Process
611
612 To keep our TASK branch in sync with the current head of tree we need
613 to do a regular merge out. the TaskMakefile contains the makefile
614 procedures to do this efficiently. NOTE: Following these procedures is
615 important in that you are merging out new material each time you do
616 the merge out. If you were just to repeatedly merge out, you would be
617 merging previously merged changes a second time causing a real mess.
618
619 Start with new directory and put TaskMakefile above pegasus (needed so you
620 have this file for the initial operations.
621
622 make -f TaskMakefile branch_merge_out BNAME=PEP317-pullop ## takes a long time
623
624 This checks out current head, merges it into task branch and sets tags
625 for the mergeout. Note that at the end of this step this work is
626 part of the TASK... branch.
627 karl 1.1.2.11
628 NOW check for conflicts, errors, etc. that resulted from the merge.
629 Look for conflict flags, compare the results (I use linux merge as a
630 good graphic compare tool) and build and test. When you are satisfied
631 that the merge out is clean, you can commit the results to the TASK...
632 branch
633
634 To commit the work to this into Task branch
635
636 make -f mak/TaskMakefile branch_merge_out_commit BNAME=PEP317-pullop
637
638 or manually commit and finish as follows
639
640 cvs commit
641 make -f mak/TaskMakefile branch_merge_out_finish BNAME=PEP317-pullop
642
643 ## This last step is important since it cleans up temporary tags to prepare
644 you for the next checkout
645
646 COMPARE TASKBRANCH WITH HEAD
647
648 karl 1.1.2.11 In a new pegasus work space do same as above for merge out.
649
650 make -f TaskMakefile BNAME=PEP317-pullop
651
652 This produces a result which is all of the head merged into the branch.
653 A diff of this is all the new changes to the head of tree that you will
654 include into the merge.
655
|
Return to
readme.pulloperations
CVS log
Up to [Pegasus] / pegasus