pegasus/readme.pulloperations - annotate

Return to readme.pulloperations CVS log

Up to [Pegasus] / pegasus

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 1.1.2.3
11 karl 1.1.2.4 TODO list:
12 karl 1.1.2.3 1. Binary operation from OOP. Need to add counter to binary 13 protocol to be able to count objects in response. Generates 14 warnings in things like messageserializer and does not work with 15 OOP right now.
16 karl 1.1.2.4 2. OpenExecQuery - Code is incorrect in that it used InstancesWithPath 17 where the spec is instances with no path. Need new function to wrap 18 getInstanceElement(withoutPathElement) in XmlReader. Note that 19 Alternate is to put flag on InstancesWith Path to say no path 20 3. Code for Pull part of OpenQueryInstancesRequest a) should be part of 21 the common CIMOperationRequestDispatcher execCommon code. 22 4. The changes to WQLCIMOperationRequestDispatcher and CQL... for handling 23 pull not completed so we feed the responses back to the EnmerationContext 24 queues
25 karl 1.1.2.3 3. Lots of minor TODOs, etc.
26 karl 1.1.2.4 4. External runtime variables. Decide this as part of PEP. The variables 27 exist in CIMOperationRequestDispatcher but not in CIMConfig. The primary 28 ones to consider are: 29 a. System maxObjectCount. Setting some maximum size on what a pull 30 client can request (i.e. the maximum size of the maxObjectCount on 31 Open... and pull operations. 32 b. Pull interoperationTimeout (max times between operations). This is 33 the maximum number of seconds on the operationTimeout parameter of the 34 Open operations 35 c. Maximum size of the responseCache before it starts backing up 36 responses to the providers.
37 karl 1.1.2.3 5. Decision on EnumerationContext timeout (separate thread or just
38 karl 1.1.2.4 checks during other operations). Can we, in fact really keep the 39 enumeration context table and queue under control without monitoring 40 with a separate thread. We must monitor for: 41 a. Client operation that stop requesting (i.e. inter operation time 42 exceeds operationTimeout). Note that if it simply exceeds the time 43 the next operation does the cleanup. The issue is those clients that 44 simply stop and do not either close or go to completion. 45 b. We should protect against providers that no not every finish delivering 46 or take to long between deliveries. This does not exist in Pegasus 47 today 48 6. Clean up code in Dispatcher. The associators code is still real mess 49 and the pull code is in a template. The Pull code is good now but 50 must be duplicated. Look at creating new CIMMessage CIMPullResponseMessage 51 so that we can have common code. Everything is the same except what 52 goes into the CIMResponseData so it is logical to have completely 53 common processing
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 1.1.2.4 8. Still using templates, etc. in code in the Dispatcher. This is for 61 all of the open operations where there is a lot of duplicate code 62 and the pull operations that are 99% duplicate code (in a single template) 63 64 Fixed for 9 June CVS update 65 1. Cleaned up code for OpenQueryInstances. Note that this is incomplete. 66 No support in WQL or CQL Operations 67 2. 68 69 What was fixed for 5 June checkin.
70 karl 1.1.2.3 1. Extended ResponseTest MOF for for both CMPI and C++ subclasses 71 2. Fixed issues with pullop. 72 3. Fixed temp issue with CIMResponseData size by putting in mutex. That 73 is not a permanent fix but it gets around issue probably in the control 74 of the move logic that meant counts were off. 75 4. Fixed issues in Dispatcher so that associator code works. Still messy 76 code in the dispatcher. 77 5. Changed name of Enumerationtable.h & cpp to EnumerationContextTable.* 78 6 Changed name of ResponseStressTest module, classes, etc. 79 80 TAG: TASK_PEP317_5JUNE_2013_2 81
82 karl 1.1.2.2 2 June 2013
83 karl 1.1.2.1 84 Issues - KS 85 1. have not installed the binary move in CIMResponseData. Please run 86 with OPP off. 87 2. Some problem in the processing so we are getting server crashes. 88 Right no I am guessing that this is in the binaryCodec and am going to 89 expand the test tools to allow testing through the localhost. 90 91 3. Still way to many TODO and KS comments and KS_TEMPS. Removing bit by bit. 92 93 4. Env variable connection for the config parameters not installed. 94 95 5. Issue with the threaded timer. For some reason during tests it 96 eventually calls the timer thread with trash for the parm (which is 97 pointer to the EnumerationTable object). Caught because we do a valid 98 test at beginning of the function. 99
100 karl 1.1.2.2 6. Still using the templates in CIMOperationRequestDispatcher to simplify 101 the handle... processing. 102 103 7. I think I have a way around the double move of objects in the 104 EnumerationContext so that the outputter will just take a defined number 105 of objects directly from the gathering cache and save the second move. 106 107 8. Not yet passing all tests but getting closer now. 108 109 9. Created a tag before this commit TASK_PEP317_1JUNE_2013. 110 111 10. Next Tag will be TASK_PEP317_2_JUNE_2013 in the task branch 112 113
114 karl 1.1.2.1 =========================================== 115 116 OVERVIEW: 117 118 The operation extensions for pull operations defined in the DMTF specification 119 DSP0200 V 1.4 were implemented in Pegasus effective Pegasus version 2.11 120 including Client and Server. 121 122 These operations extend the CIM/XML individual operations to operation 123 sequences where the server must maintain state between operations in a 124 sequence and the client must execute multiple operations to get the full 125 set of instances or instance paths. 126 127 The following new CIM/XML operations as defined in DSP0200 are included; 128 129 -OpenEnumerateInstances 130 -openEnumerateInstancePaths 131 -OpenReferenceInstances 132 -OpenReferenceInstancePaths 133 -OpenAssociatiorInstances 134 -OpenAssociatorInstancePaths 135 karl 1.1.2.1 -PullInstancesWithPath 136 -PullInstancePaths 137 -CloseEnumeration 138 -EnumerationCount
139 karl 1.1.2.2 OpenExecQuery
140 karl 1.1.2.1 141 The following operations have not been implemented in this version of Pegasus: 142 143 -OpenQueryInstances 144 145 The following limitations on the implementation exist; 146 147 1. The filterQueryLanguage and filterQuery parameters are processed by 148 the Pegasus client but the server returns error if there is any data in
149 karl 1.1.2.2 either parameter. This work does not include the development of the 150 query language. Note that a separate effort to extend Pegasus to use 151 the DMTF FQL query language is in process.
152 karl 1.1.2.1 153 2. The input parameter continueOnError is processed correctly by the client 154 but the Pegasus server only provides for false since the server does not 155 include logic to continue processing responses after an error is 156 encountered. 157 This is consistent with the statement in the specification that use of 158 this functionality is optional and the fact that the DMTF agrees that all 159 of the issues of continuing after errors have not been clarified. 160 161 3. The operation enumerationCount is not processed by the server today since 162 a) really getting the count would be the same cost as the corresponding 163 enumeration, b) the server does not include a history or estimating 164 mechanism for this to date.
165 karl 1.1.2.2 NOTE: After a through review as part of the development of the next version 166 of CMPI we have concluded that this operation is probably not worth the 167 effort. Since it is optional, Pegasus will only return the unknown status 168 at this point
169 karl 1.1.2.1 170 Since the concept of sequences of operations linked together (open, pull, close) 171 is a major extension to the original CIM/XML operation concept of completely 172 independent operations several new pieces of functionality are implemented 173 to control interOperationTimeouts, counts of objects to be returned, etc. 174 175 TBD - Review this 176 177 CLIENT 178 179 The new operations follow the same pattern as the APIs for existing operations 180 in that: 181 182 1. All errors are handled as CIMException and Exception 183 184 2. The means of inputting parameters are the same except that there are 185 significantly more input parameters with the open operations and for the 186 first time operations return parameters as well as objects in the 187 response. Specifically the open and pull operations return values for 188 enumerationContext which is the identity for a pull sequence and 189 endOfSequence which is the marker the server sends in open and pull 190 karl 1.1.2.1 responses when it has no more objects to send. 191 192 The significant differences include: 193 194 1. Processing of parameters on responses (i.e. the endOfSequence and 195 enumerationContext parameters are returned for open and pull operations). 196 197 2. Numeric arguments (Uint32 and Uint64 include the option of NULL in some 198 cases so they are packaged inside classes Uint32Arg and Uint64Arg in the 199 client api. 200 201 3. The association and reference operations ONLY process instances. They do 202 not include the capability to return classes like reference and associator 203 do and therefore return CIMInstance rather than CIMObject. 204 205 4. Paths are returned in all cases (i.e OpenEnumerateInstances and 206 PullInstancesWithPath where they were not with EnumeratInstances. 207 208 5. The client must maintain state between operations in a sequence (using 209 the enumerationContext parameter). 210 211 karl 1.1.2.1 TBD- Are there more differences. 212 213 214 SERVER 215 216 The Pegasus server attempts to always deliver the requested number of objects 217 for any open or pull request (the specification allows for the server to 218 deliver less than the requested number of objects and specifically to return 219 zero objects on open). We felt that it was worth any extra cost in processing 220 to provide the client with exactly what it had requested. 221 222 The pegasus server always closes an enumeration sequence upon receipt of any 223 error from the providers, repository, etc. Therefore the server will reject
224 karl 1.1.2.2 any request that has continueOnError = true; 225 226 Expansion to allow the continue on error may be added in a future version. 227 In any case, the whole purpose of the continue on error is really to allow 228 input from good providers to be mixed with providers that return errors so 229 that generally this would mean simply changing the logic in the return mechanism 230 to not shutdown when an error is received from any given provider. 231 232 Generally we do not believe that the providers need to do much more in the 233 future to support the continueOnError other than possibly allowing the provider 234 to continue processing after it has received an error.
235 karl 1.1.2.1 236 PROVIDERS 237 238 This implementation requires NO changes to the existing providers. The 239 provider APIs operate just as they do with the original operations. 240 241 Because the server processing is different however, there may be some 242 behavior differences primarily because the client now controls the speed of 243 delivery of objects. 244 245 In previous versions of Pegasus, the server attempts to deliver objects as 246 rapidly as then can be put on the network. In the case of HTTP chunked requests 247 they are delivered in chunks of about 100 objects. The primary delay for the 248 providers was the processing of each segment through the server. The server 249 is blocked so that no other segment can proceed through the server until that 250 segment is processed and sent on the network. 251 In the case of non-chunkedresponses, they are completely gathered in the serve 252 and then delivered as one non-chunked response. There were no delays for the 253 providers, just lots of possible memory use in the server. 254 255 The responses from providers (delivered through the deliver(...) interface are 256 karl 1.1.2.1 gathered into segments of about 100 objects and this group of objects is moved 257 through the server to be delivered to the client. 258 259 However with the inclusion of the pull operations, The segments of objects 260 from the providers are cached in the server response path until the 261 maxObjectCount for that request (open or pull) and that number returned in a 262 non-chunked response. Thus, if the client is slow to issue pull requests, 263 the providers might be delayed at some point to reduce memory usage in the 264 server (the delay appears as slow response tothe deliver operation). 265 266 In other words, the time to process large sets of responses from the provider 267 now depends on the speed of handling the client. 268 269 It is important to remember in developing providers that the Pegasus server 270 can most efficiently process responses if they are passed from the provider 271 to the server individually or in small arrays of objects rather than the 272 provider gathering very large arrays of objects and sending them to the 273 server. 274
275 karl 1.1.2.2 NEXT GENERATION PROVIDERS 276 KS_TODO 277
278 karl 1.1.2.1 CONFIGURATION PARAMETERS 279 280 The server includes several configuration parameters to set limits on the 281 processing of pull operations. All of these configuration parameters are 282 compile time parameters rather than runtime. 283 284 1. Maximum value of minimum interoperation time. This parameter defines the 285 maximum time allowed between the return of an open or pull response and 286 the receipt of the next pull or a close operation before the server may 287 close the enumeration. The specification allows the server to set a 288 maximum interoperation time and refuse open requests that with requested 289 operationTimeout greater than that time. 290 CIM_ERR_INVALID_OPERATION_TIMEOUT 291 292 This value is set with the Pegasus environment variable 293 PEGASUS_PULL.... 294 295 2. Maximum objects returned in a single open or pull operation. The server 296 can set a maximum limit on the number of objects that can be returned in 297 a single open or pull oepration with the maxObjectCount parameter. 298 299 karl 1.1.2.1 3. Whether the server allows 0 as an interoperation timeout value. The value 300 zero is s special value for the interoperationTimeout in that it tells the 301 server to not timeout any enumeration sequence. 302 303 With this value for interoperationTimeout, the only way to close an 304 enumeration sequence is to complete all of the pulls or issue the close. 305 If for some reason the sequence is not completed, that enumeration context 306 would remain open indefinitly. Since in Pegasus any open enumeration 307 context uses resources (the context object and any provider resposnes that 308 have not yet been issued in a response) it would appear that most 309 platforms would not want to allow the existence of enumeration contexts 310 that cannot be closed by the server. 311 312 4, maximum consecutive pull requests with 0 maxObjectCount. The use of the 313 pull operation with maxObjectCount set to zero could be used to keep an 314 enumeration context open indefinitly (this tells the server to restart the 315 interoperationTimeout but not send any objects in the response). Therefore the 316 specification allows for the server setting maximum limits on this behavior 317 and returning the error CIM_ERR_SERVER_LIMITS_EXCEEDED if this limit is 318 exceeded. 319 Note that this is maximum CONSECUTIVE pulls so that issuing a pull with 320 karl 1.1.2.1 a non-zero count resets this counter. 321 322 KS-TBD - Is this really logical since we can still block by just issuing 323 lots of zero request and an occansional request for one object. 324 325 Pegaus sets the value of this limit to 1000 and allows the implementer to 326 modify it with the PEGASUS_MAXIMUM_ZERO_OBJECTCOUNT environment variable. 327 328 5. Default operationTimeout - 329 330 The default of this parameter is to refuse operat 331 332 In the current release of Pegasus these are all compile time parameters.

No CVS admin address has been configured