Superfile with missing subfile

[DSC]

I'm building a set of superfiles and subfiles to be used in a data update scenario. In working through the ECL I've made several mistakes, and I now have a peculiar situation. I have a superfile that, according to the 'Browse Logical Files' screen contains one subfile but that subfile doesn't actually exist. Furthermore, when trying to either delete the superfile or to even view its details, I get the following error:

Code: Select all

Reporter: WsDfu::DFUARRAYACTION()

-1   2012-09-24 15:53:22 GMT: CDistributedSuperFile: SuperFile poc::test_data::data is missing sub-file file poc::test_data::update_data

How can I get rid of this superfile?

Thanks,

Dan

[DSC]

It turns out that File.DeleteSuperFile() does remove the superfile from the logical file list, as long as I don't pass TRUE for the second parameter ('delete subfiles'). I don't know if this leaves the file system in an inconsistent state, though. Any information along those lines would be appreciated.

Cheers,

Dan

[bforeman]

Hi Dan,

Your file system should still be rock solid. I believe the issue is in the Delete function of the ECL Watch. We are currently investigating this issue as you always have been able to Delete a Superfile in the ECL Watch, even if the sub file had been removed, but we just noticed this in the most recent 3.8 version.

Regards,

Bob

[DSC]

Well, now I've run into a superfile that refuses to be deleted, even through code. This particular superfile believes that it contains a sub-superfile that itself contains data. Executing File.DeleteSuperFile() now results in a segfault:

Code: Select all

00000005 2012-09-24 13:14:16  6991  6992 AgentExec: Executing 'start_eclagent WUID=W20120924-131415 DALISERVERS=10.210.150.113:7070'
00000006 2012-09-24 13:14:16  6991  6992 Execution started
00000007 2012-09-24 13:14:16  6991  6992 AgentExec: Waiting on queue(s) 'hthor.agent,thor.agent'
00000000 2012-09-24 13:14:16  7580  7580 Logging to /var/log/HPCCSystems/myeclagent/eclagent.2012_09_24.log
00000001 2012-09-24 13:14:16  7580  7580 ECLAGENT build community_3.8.2-1
00000002 2012-09-24 13:14:16  7580  7580 Waiting for workunit lock
00000003 2012-09-24 13:14:16  7580  7580 Obtained workunit lock
00000004 2012-09-24 13:14:17  7580  7580 Loading dll (libW20120924-131415.so) from location /var/lib/HPCCSystems/myeclccserver/libW20120924-131415.so
00000005 2012-09-24 13:14:17  7580  7580 Starting process
00000006 2012-09-24 13:14:17  7580  7580 RoxieMemMgr: Setting memory limit to 314572800 bytes (300 pages)
00000007 2012-09-24 13:14:17  7580  7580 RoxieMemMgr: 320 Pages successfully allocated for the pool - memsize=335544320 base=0x2aaaab600000 alignment=1048576 bitmapSize=10
00000008 2012-09-24 13:14:17  7580  7580 Waiting for run lock
00000009 2012-09-24 13:14:17  7580  7580 Obtained run lock
0000000A 2012-09-24 13:14:17  7580  7580 CDistributedSuperFile: SuperFile test_poc::test_data::all_data is missing sub-file file test_poc::test_data::update_data
0000000B 2012-09-24 13:14:17  7580  7580 ================================================
0000000C 2012-09-24 13:14:17  7580  7580 Signal:    11 Segmentation fault
0000000D 2012-09-24 13:14:17  7580  7580 Fault IP:  00002B80756F1A0F
0000000E 2012-09-24 13:14:17  7580  7580 Accessing: 0000000000000000
0000000F 2012-09-24 13:14:17  7580  7580 Registers:
00000010 2012-09-24 13:14:17  7580  7580 EAX:0000000017191A4C  EBX:0000000000000000  ECX:0000000017191A40  EDX:0000000000000000  ESI:0000000000000000  EDI:0000000000000000
00000011 2012-09-24 13:14:17  7580  7580 CS:EIP:0033:00002B80756F1A0F
00000012 2012-09-24 13:14:17  7580  7580    ESP:00007FFF439ACCF0  EBP:000000001719221C
00000013 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACCF0]: 0000000000001D9C 73FAC9D800000000 00002B8073FAC9D8 0000000100002B80 0000000000000001 7454FAC300000000 00002B807454FAC3 1719142000002B80
00000014 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACD10]: 0000000017191420 439ACDB000000000 00007FFF439ACDB0 439ACDA000007FFF 00007FFF439ACDA0 1719177800007FFF 0000000017191778 439ACD6000000000
00000015 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACD30]: 00007FFF439ACD60 0000271000007FFF 01002B8000002710 1719172001002B80 0000000017191720 FFFFFFFF00000000 00000001FFFFFFFF 1719169800000001
00000016 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACD50]: 0000000017191698 0037D46000000000 000000000037D460 171914A000000000 00000000171914A0 0000001600000000 0000000000000016 0000000000000000
00000017 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACD70]: 0000000000000000 0000000000000000 0000000000000000 7448000000000000 00002B8074480000 0000000000002B80 0000000000000000 171930B000000000
00000018 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACD90]: 00000000171930B0 0000007A00000000 000008000000007A 0000000200000800 0000000000000002 1718CD1000000000 000000001718CD10 1719180000000000
00000019 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACDB0]: 0000000017191800 0000002100000000 0000004000000021 0000000000000040 0000000000000000 0000000000000000 0000000000000000 0000000000000000
0000001A 2012-09-24 13:14:17  7580  7580 Stack[00007FFF439ACDD0]: 0000000000000000 0000271000000000 0000000000002710 171910D000000000 00000000171910D0 0000000100000000 0000000000000001 1719168000000000
0000001B 2012-09-24 13:14:17  7580  7580 Backtrace:
0000001C 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libjlib.so(_Z16PrintStackReportv+0x26) [0x2b80756a6e06]
0000001D 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libjlib.so(_Z13excsighandleriP7siginfoPv+0x295) [0x2b80756a7e55]
0000001E 2012-09-24 13:14:17  7580  7580   /lib64/libpthread.so.0 [0x3d4bc0eb70]
0000001F 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libjlib.so(_Z18createPTreeFromIPTPK13IPropertyTree9ipt_flags+0x1f) [0x2b80756f1a0f]
00000020 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libdalibase.so(_ZN21CDistributedSuperFile12loadSubFilesEbP27IDistributedFileTransactionj+0x353) [0x2b807454fac3]
00000021 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libdalibase.so(_ZN21CDistributedSuperFileC1EP25CDistributedFileDirectoryP17IRemoteConnectionRK19CDfsLogicalFileNameP15IUserDescriptorbP27IDistributedFileTransactionbj+0xf4) [0x2b807455bc24]
00000022 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libdalibase.so(_ZN25CDistributedFileDirectory8dolookupERK19CDfsLogicalFileNameP15IUserDescriptorbP27IDistributedFileTransactionbj+0x165) [0x2b80745423d5]
00000023 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libdalibase.so(_ZN25CDistributedFileDirectory15lookupSuperFileEPKcP15IUserDescriptorP27IDistributedFileTransactionbj+0x6e) [0x2b807454286e]
00000024 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libdalibase.so(_ZN27CDistributedFileTransaction15lookupSuperFileEPKcbj+0x95) [0x2b8074553f95]
00000025 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/plugins/libfileservices.so [0x2aaaaae74f52]
00000026 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/plugins/libfileservices.so(fsDeleteSuperFile+0x65) [0x2aaaaae76545]
00000027 2012-09-24 13:14:17  7580  7580   /var/lib/HPCCSystems/myeclccserver/libW20120924-131415.so [0x2aaaab3e8fb6]
00000028 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libeclrtl.so(_ZN15WorkflowMachine11performItemEjj+0x54) [0x2b80751a31c4]
00000029 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libeclrtl.so(_ZN15WorkflowMachine13doExecuteItemER20IRuntimeWorkflowItemj+0x3f) [0x2b80751a3b3f]
0000002A 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libeclrtl.so(_ZN15WorkflowMachine11executeItemEjj+0x26a) [0x2b80751a35da]
0000002B 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libeclrtl.so(_ZN15WorkflowMachine7performEP18IGlobalCodeContextP11IEclProcess+0x139) [0x2b80751a41b9]
0000002C 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libhthor.so(_ZN8EclAgent10runProcessEP11IEclProcess+0x14a) [0x2b8072aa503a]
0000002D 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libhthor.so(_ZN8EclAgent9doProcessEv+0x283) [0x2b8072aa6fd3]
0000002E 2012-09-24 13:14:17  7580  7580   /opt/HPCCSystems/lib/libhthor.so(_Z13eclagent_mainiPPKcP12StringBufferb+0x6af) [0x2b8072aabf0f]
0000002F 2012-09-24 13:14:17  7580  7580   eclagent(main+0x61) [0x4011a1]
00000030 2012-09-24 13:14:17  7580  7580   /lib64/libc.so.6(__libc_start_main+0xf4) [0x3d4b01d994]
00000031 2012-09-24 13:14:17  7580  7580   eclagent(__gxx_personality_v0+0xe9) [0x401079]
00000032 2012-09-24 13:14:17  7580  7580 ThreadList:
439D1940 1134369088 7581: CMPNotifyClosedThread
459D2940 1167927616 7582: MP Connection Thread
499D4940 1235044672 7584: CSocketSelectThread
479D3940 1201486144 7585: LogMsgParentReceiver
4B9D5940 1268603200 7587: LogMsgFilterReceiver
4D9D6940 1302161728 7588: EclAgent Abort Monitor
4F9D7940 1335720256 7589: CDaliPublisherClient

00000033 2012-09-24 13:14:17  7580  7580 SIG: Segmentation fault(11), accessing 0000000000000000, IP=00002B80756F1A0F

test_poc::test_data::update_data (the superfile that is the referenced subfile) did exist at one time.

Is there any way to clear this thing from my list?

Cheers,

Dan

[bforeman]

Hi Dan,

I have an issue opened with the development team, and one of us will reply back as soon as we have a solution. Meanwhile simply ignore that file on your cluster, it is simply an entry and is not taking up any resources.

Regards,

Bob

[rtaylor]

Dan,

You can maintain superfiles in a GUI interface through ECL Watch, too. So you can open the sub-superfile's Logical File Details page and remove any sub-file references there, too.

It also might help to know that superfiles themselves don't really exist!
They are simply meta-data in the DFU that are treated as if they were actual logical files.

HTH,

Richard

[DSC]

Huh.

When I had the problem earlier, before finding the workaround with File.DeleteSuperfile(), clicking on the Details option for the superfile generated an error rather than the superfile details. This time, when File.DeleteSuperfile() segfaults, Details works and I can delete everything without a problem.

Go figure.

I guess the take-away here is to always get you guys involved, burning time and bandwidth, and the solution will just present itself.

Cheers!

Dan

[bforeman]

Dan,

I received a message this morning that this issue has been fixed in the next update. Thanks again for your report.

Regards,

Bob