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 OOP right now.
|
15 karl 1.1.2.4 2. OpenExecQuery - Code is incorrect in that it used InstancesWithPath
16 where the spec is instances with no path. Need new function to wrap
17 getInstanceElement(withoutPathElement) in XmlReader. Note that
18 Alternate is to put flag on InstancesWith Path to say no path
19 3. Code for Pull part of OpenQueryInstancesRequest a) should be part of
20 the common CIMOperationRequestDispatcher execCommon code.
21 4. The changes to WQLCIMOperationRequestDispatcher and CQL... for handling
22 pull not completed so we feed the responses back to the EnmerationContext
23 queues
|
24 karl 1.1.2.7 3. Lots of minor TODOs, diagnostics, etc.
|
25 karl 1.1.2.4 4. External runtime variables. Decide this as part of PEP. The variables
26 exist in CIMOperationRequestDispatcher but not in CIMConfig. The primary
27 ones to consider are:
28 a. System maxObjectCount. Setting some maximum size on what a pull
29 client can request (i.e. the maximum size of the maxObjectCount on
30 Open... and pull operations.
31 b. Pull interoperationTimeout (max times between operations). This is
32 the maximum number of seconds on the operationTimeout parameter of the
33 Open operations
34 c. Maximum size of the responseCache before it starts backing up
35 responses to the providers.
|
36 karl 1.1.2.3 5. Decision on EnumerationContext timeout (separate thread or just
|
37 karl 1.1.2.4 checks during other operations). Can we, in fact really keep the
38 enumeration context table and queue under control without monitoring
39 with a separate thread. We must monitor for:
40 a. Client operation that stop requesting (i.e. inter operation time
41 exceeds operationTimeout). Note that if it simply exceeds the time
42 the next operation does the cleanup. The issue is those clients that
43 simply stop and do not either close or go to completion.
44 b. We should protect against providers that no not every finish delivering
45 or take to long between deliveries. This does not exist in Pegasus
46 today
|
47 karl 1.1.2.8 6. Clean up code in Dispatcher. Want to at least reduce the code for the
48 Open Operations to a set of templates so we know that the code is the
49 same for all operations. Right now the existing operations enum, assoc,
50 etc. are all in templates but not the open operations. Note that the
51 pull is a single template for both pullInstances and pullInstancePaths.
52 However, we might be able to reduce this to a single function by adding
53 a new level to CIMMessage.h (CommonPullResponse)
|
54 karl 1.1.2.3 7. Extension to avoid double move of objects in CIMResponseData (one
55 into enumerationContext queue and second to new cimResponseData for
56 response. Want to avoid second move by extending Open/Pull response
57 messages to include count and CIMResponse data to count objects out
58 of queue when converting (avoids the second move). Big issue here
59 with binary data since need to extend format to count it.
|
60 karl 1.1.2.8 8. NEXT TASKS:
61 a. get the pull operations rather than a template into a single
62 function by creating a new CIMPullResponse message in CIMMessage.h that
63 contains the pull data. Then we can use a single function to process all
64 pull operations.
|
65 karl 1.1.2.9 b. test the enumeration timeout thread
|
66 karl 1.1.2.8 c. New Mergout to bring up to to date again.
|
67 karl 1.1.2.11 d. Sort out issues with statisticalData since it really does not work
68 with Open and Pull operations
|
69 karl 1.1.2.10
70 29 September 2013. CVS update.
71 1. Modified calls to statisticalData.cpp to a) directly call with request
72 type, b) incorporate the open, pull, etc. messages. However, since these
73 are not part of the CIM class, we must do something special with them.
74 2. Corrected OOP interface to enable new flag to indicate internal operations
75 and set host, etc.
76 3. Add code to CQLOperationsDispatcher and WQLOperationDispatcher to clean
77 up CIMResponseDataCounter after filtering.
78 4. Modified ProviderAgent to set Host info for some pull operations.
79 5. Added new flag to CIMBinMsgSerializer and Deserializer.
|
80 karl 1.1.2.8
|
81 karl 1.1.2.9 17 September 2013 CVS update (Actually two different updates over 3 days)
82 1. Clean up some issues in CIMMessage.h and CIMMessage.cpp
83 2. Extend OpenExecQuery to WQL and CQL processors but return not complete
84 3. Remove memory leak in EnumerationContext and EnumerationContextTable
85 handling.
86 4. Created template functions for much of the pull operations.
87 5. Reversed order of queryLanguage and query (and changed names to match
88 execQuery) in client and server. Note that these are the execQuery
89 WQL and CQL filters and NOT FQL filters.
90 6. Some code cleanup in dispatcher
91 7. Today, passes all tests in pullop but issue in alltests. For some reason
92 not finding CIMObjectManager instance. Also, leaves enumeration contexts
93 if client terminates since cleanup thread not operating.
|
94 karl 1.1.2.11 8. XML from OOP not correctly processed.
|
95 karl 1.1.2.9
|
96 karl 1.1.2.8 14 September 2013 CVS update
97 Merged out up to 25 August. Cleaned up all operations and standardized code.
98 At this point the non pull operations code is in a set of templates but the
99 pull is not yet.
100 Fixed a significant number of problems so that it appears that the operations
101 except for OpenExecQuery run stably, at least with the pullop test program.
102 Note that there is a problem in that the Interop control provider is not
103 returning its singleton wbemserver object for some reason. Causes a test
104 failure
|
105 karl 1.1.2.7
106 Fixed for 16 June CVS Update
107 1. Cleaned up the enumerationContext and Table release functions and tested
108 to confirm that we do not lose memory in either normal sequences or
109 sequences that close early. Cleaned up pullop and added more tests
|
110 karl 1.1.2.8 Taged Before: PREAUG25UPDATE and after POSTAUG25UPDATE
|
111 karl 1.1.2.4
112 Fixed for 9 June CVS update
113 1. Cleaned up code for OpenQueryInstances. Note that this is incomplete.
114 No support in WQL or CQL Operations
115 2.
116
117 What was fixed for 5 June checkin.
|
118 karl 1.1.2.3 1. Extended ResponseTest MOF for for both CMPI and C++ subclasses
119 2. Fixed issues with pullop.
120 3. Fixed temp issue with CIMResponseData size by putting in mutex. That
121 is not a permanent fix but it gets around issue probably in the control
122 of the move logic that meant counts were off.
123 4. Fixed issues in Dispatcher so that associator code works. Still messy
124 code in the dispatcher.
125 5. Changed name of Enumerationtable.h & cpp to EnumerationContextTable.*
126 6 Changed name of ResponseStressTest module, classes, etc.
127
128 TAG: TASK_PEP317_5JUNE_2013_2
129
|
130 karl 1.1.2.2 2 June 2013
|
131 karl 1.1.2.1
132 Issues - KS
133 1. have not installed the binary move in CIMResponseData. Please run
134 with OPP off.
135 2. Some problem in the processing so we are getting server crashes.
136 Right no I am guessing that this is in the binaryCodec and am going to
137 expand the test tools to allow testing through the localhost.
138
139 3. Still way to many TODO and KS comments and KS_TEMPS. Removing bit by bit.
140
141 4. Env variable connection for the config parameters not installed.
142
143 5. Issue with the threaded timer. For some reason during tests it
144 eventually calls the timer thread with trash for the parm (which is
145 pointer to the EnumerationTable object). Caught because we do a valid
146 test at beginning of the function.
147
|
148 karl 1.1.2.2 6. Still using the templates in CIMOperationRequestDispatcher to simplify
149 the handle... processing.
150
151 7. I think I have a way around the double move of objects in the
152 EnumerationContext so that the outputter will just take a defined number
153 of objects directly from the gathering cache and save the second move.
154
155 8. Not yet passing all tests but getting closer now.
156
157 9. Created a tag before this commit TASK_PEP317_1JUNE_2013.
158
159 10. Next Tag will be TASK_PEP317_2_JUNE_2013 in the task branch
160
161
|
162 karl 1.1.2.1 ===========================================
163
164 OVERVIEW:
165
166 The operation extensions for pull operations defined in the DMTF specification
167 DSP0200 V 1.4 were implemented in Pegasus effective Pegasus version 2.11
168 including Client and Server.
169
170 These operations extend the CIM/XML individual operations to operation
171 sequences where the server must maintain state between operations in a
172 sequence and the client must execute multiple operations to get the full
173 set of instances or instance paths.
174
175 The following new CIM/XML operations as defined in DSP0200 are included;
176
177 -OpenEnumerateInstances
178 -openEnumerateInstancePaths
179 -OpenReferenceInstances
180 -OpenReferenceInstancePaths
181 -OpenAssociatiorInstances
182 -OpenAssociatorInstancePaths
183 karl 1.1.2.1 -PullInstancesWithPath
184 -PullInstancePaths
185 -CloseEnumeration
186 -EnumerationCount
|
187 karl 1.1.2.2 OpenExecQuery
|
188 karl 1.1.2.1
189 The following operations have not been implemented in this version of Pegasus:
190
191 -OpenQueryInstances
192
193 The following limitations on the implementation exist;
194
195 1. The filterQueryLanguage and filterQuery parameters are processed by
196 the Pegasus client but the server returns error if there is any data in
|
197 karl 1.1.2.2 either parameter. This work does not include the development of the
198 query language. Note that a separate effort to extend Pegasus to use
199 the DMTF FQL query language is in process.
|
200 karl 1.1.2.1
201 2. The input parameter continueOnError is processed correctly by the client
202 but the Pegasus server only provides for false since the server does not
203 include logic to continue processing responses after an error is
204 encountered.
205 This is consistent with the statement in the specification that use of
206 this functionality is optional and the fact that the DMTF agrees that all
207 of the issues of continuing after errors have not been clarified.
208
209 3. The operation enumerationCount is not processed by the server today since
210 a) really getting the count would be the same cost as the corresponding
211 enumeration, b) the server does not include a history or estimating
212 mechanism for this to date.
|
213 karl 1.1.2.2 NOTE: After a through review as part of the development of the next version
214 of CMPI we have concluded that this operation is probably not worth the
215 effort. Since it is optional, Pegasus will only return the unknown status
216 at this point
|
217 karl 1.1.2.1
218 Since the concept of sequences of operations linked together (open, pull, close)
219 is a major extension to the original CIM/XML operation concept of completely
220 independent operations several new pieces of functionality are implemented
221 to control interOperationTimeouts, counts of objects to be returned, etc.
222
223 TBD - Review this
224
225 CLIENT
226
227 The new operations follow the same pattern as the APIs for existing operations
228 in that:
229
230 1. All errors are handled as CIMException and Exception
231
232 2. The means of inputting parameters are the same except that there are
233 significantly more input parameters with the open operations and for the
234 first time operations return parameters as well as objects in the
235 response. Specifically the open and pull operations return values for
236 enumerationContext which is the identity for a pull sequence and
237 endOfSequence which is the marker the server sends in open and pull
238 karl 1.1.2.1 responses when it has no more objects to send.
239
240 The significant differences include:
241
242 1. Processing of parameters on responses (i.e. the endOfSequence and
243 enumerationContext parameters are returned for open and pull operations).
244
245 2. Numeric arguments (Uint32 and Uint64 include the option of NULL in some
246 cases so they are packaged inside classes Uint32Arg and Uint64Arg in the
247 client api.
248
249 3. The association and reference operations ONLY process instances. They do
250 not include the capability to return classes like reference and associator
251 do and therefore return CIMInstance rather than CIMObject.
252
253 4. Paths are returned in all cases (i.e OpenEnumerateInstances and
254 PullInstancesWithPath where they were not with EnumeratInstances.
255
256 5. The client must maintain state between operations in a sequence (using
257 the enumerationContext parameter).
258
259 karl 1.1.2.1 TBD- Are there more differences.
260
261
262 SERVER
263
264 The Pegasus server attempts to always deliver the requested number of objects
265 for any open or pull request (the specification allows for the server to
266 deliver less than the requested number of objects and specifically to return
267 zero objects on open). We felt that it was worth any extra cost in processing
268 to provide the client with exactly what it had requested.
269
270 The pegasus server always closes an enumeration sequence upon receipt of any
271 error from the providers, repository, etc. Therefore the server will reject
|
272 karl 1.1.2.2 any request that has continueOnError = true;
273
274 Expansion to allow the continue on error may be added in a future version.
275 In any case, the whole purpose of the continue on error is really to allow
276 input from good providers to be mixed with providers that return errors so
277 that generally this would mean simply changing the logic in the return mechanism
278 to not shutdown when an error is received from any given provider.
279
280 Generally we do not believe that the providers need to do much more in the
281 future to support the continueOnError other than possibly allowing the provider
282 to continue processing after it has received an error.
|
283 karl 1.1.2.1
284 PROVIDERS
285
286 This implementation requires NO changes to the existing providers. The
287 provider APIs operate just as they do with the original operations.
288
289 Because the server processing is different however, there may be some
290 behavior differences primarily because the client now controls the speed of
291 delivery of objects.
292
293 In previous versions of Pegasus, the server attempts to deliver objects as
294 rapidly as then can be put on the network. In the case of HTTP chunked requests
295 they are delivered in chunks of about 100 objects. The primary delay for the
296 providers was the processing of each segment through the server. The server
297 is blocked so that no other segment can proceed through the server until that
298 segment is processed and sent on the network.
299 In the case of non-chunkedresponses, they are completely gathered in the serve
300 and then delivered as one non-chunked response. There were no delays for the
301 providers, just lots of possible memory use in the server.
302
303 The responses from providers (delivered through the deliver(...) interface are
304 karl 1.1.2.1 gathered into segments of about 100 objects and this group of objects is moved
305 through the server to be delivered to the client.
306
307 However with the inclusion of the pull operations, The segments of objects
308 from the providers are cached in the server response path until the
309 maxObjectCount for that request (open or pull) and that number returned in a
310 non-chunked response. Thus, if the client is slow to issue pull requests,
311 the providers might be delayed at some point to reduce memory usage in the
312 server (the delay appears as slow response tothe deliver operation).
313
314 In other words, the time to process large sets of responses from the provider
315 now depends on the speed of handling the client.
316
317 It is important to remember in developing providers that the Pegasus server
318 can most efficiently process responses if they are passed from the provider
319 to the server individually or in small arrays of objects rather than the
320 provider gathering very large arrays of objects and sending them to the
321 server.
322
|
323 karl 1.1.2.2 NEXT GENERATION PROVIDERS
324 KS_TODO
325
|
326 karl 1.1.2.1 CONFIGURATION PARAMETERS
327
328 The server includes several configuration parameters to set limits on the
329 processing of pull operations. All of these configuration parameters are
330 compile time parameters rather than runtime.
331
332 1. Maximum value of minimum interoperation time. This parameter defines the
333 maximum time allowed between the return of an open or pull response and
334 the receipt of the next pull or a close operation before the server may
335 close the enumeration. The specification allows the server to set a
336 maximum interoperation time and refuse open requests that with requested
337 operationTimeout greater than that time.
338 CIM_ERR_INVALID_OPERATION_TIMEOUT
339
340 This value is set with the Pegasus environment variable
341 PEGASUS_PULL....
342
343 2. Maximum objects returned in a single open or pull operation. The server
344 can set a maximum limit on the number of objects that can be returned in
345 a single open or pull oepration with the maxObjectCount parameter.
346
347 karl 1.1.2.1 3. Whether the server allows 0 as an interoperation timeout value. The value
348 zero is s special value for the interoperationTimeout in that it tells the
349 server to not timeout any enumeration sequence.
350
351 With this value for interoperationTimeout, the only way to close an
352 enumeration sequence is to complete all of the pulls or issue the close.
353 If for some reason the sequence is not completed, that enumeration context
354 would remain open indefinitly. Since in Pegasus any open enumeration
355 context uses resources (the context object and any provider resposnes that
356 have not yet been issued in a response) it would appear that most
357 platforms would not want to allow the existence of enumeration contexts
358 that cannot be closed by the server.
359
360 4, maximum consecutive pull requests with 0 maxObjectCount. The use of the
361 pull operation with maxObjectCount set to zero could be used to keep an
362 enumeration context open indefinitly (this tells the server to restart the
363 interoperationTimeout but not send any objects in the response). Therefore the
364 specification allows for the server setting maximum limits on this behavior
365 and returning the error CIM_ERR_SERVER_LIMITS_EXCEEDED if this limit is
366 exceeded.
367 Note that this is maximum CONSECUTIVE pulls so that issuing a pull with
368 karl 1.1.2.1 a non-zero count resets this counter.
369
370 KS-TBD - Is this really logical since we can still block by just issuing
371 lots of zero request and an occansional request for one object.
372
373 Pegaus sets the value of this limit to 1000 and allows the implementer to
374 modify it with the PEGASUS_MAXIMUM_ZERO_OBJECTCOUNT environment variable.
375
376 5. Default operationTimeout -
377
378 The default of this parameter is to refuse operat
379
380 In the current release of Pegasus these are all compile time parameters.
|
381 karl 1.1.2.11
382
383 NOTES On working with task branch.
384
385 Merge out Process
386
387 To keep our TASK branch in sync with the current head of tree we need
388 to do a regular merge out. the TaskMakefile contains the makefile
389 procedures to do this efficiently. NOTE: Following these procedures is
390 important in that you are merging out new material each time you do
391 the merge out. If you were just to repeatedly merge out, you would be
392 merging previously merged changes a second time causing a real mess.
393
394 Start with new directory and put TaskMakefile above pegasus (needed so you
395 have this file for the initial operations.
396
397 make -f TaskMakefile branch_merge_out BNAME=PEP317-pullop ## takes a long time
398
399 This checks out current head, merges it into task branch and sets tags
400 for the mergeout. Note that at the end of this step this work is
401 part of the TASK... branch.
402 karl 1.1.2.11
403 NOW check for conflicts, errors, etc. that resulted from the merge.
404 Look for conflict flags, compare the results (I use linux merge as a
405 good graphic compare tool) and build and test. When you are satisfied
406 that the merge out is clean, you can commit the results to the TASK...
407 branch
408
409 To commit the work to this into Task branch
410
411 make -f mak/TaskMakefile branch_merge_out_commit BNAME=PEP317-pullop
412
413 or manually commit and finish as follows
414
415 cvs commit
416 make -f mak/TaskMakefile branch_merge_out_finish BNAME=PEP317-pullop
417
418 ## This last step is important since it cleans up temporary tags to prepare
419 you for the next checkout
420
421 COMPARE TASKBRANCH WITH HEAD
422
423 karl 1.1.2.11 In a new pegasus work space do same as above for merge out.
424
425 make -f TaskMakefile BNAME=PEP317-pullop
426
427 This produces a result which is all of the head merged into the branch.
428 A diff of this is all the new changes to the head of tree that you will
429 include into the merge.
430
|