[Info-ingres] Vector giving me the shits

Tue Jul 25 09:01:27 UTC 2017

Hi Dean,

I've got a nice test case and put in place the extra logging you have mentioned. The result in vectorwise.log is now:
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:Segmentation fault detected (SIGSEGV). System shutting down.
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:X100 Server version: x100-gcc-OPTXPROF-64-r12350 (branches/vw5.0) built on Sep 16 2016 15:27:40 with GCC 4.9.1
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:SIGSEGV trace:
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100(sig_backtrace_handler+0xc0)[0x5e98d4]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100(server_quit+0x93)[0x641f5f]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:/lib64/libpthread.so.0(+0xf130)[0x7fb25c768130]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100(aggr_switch2ordered+0x19a4)[0x1007a28]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100(AggrBuild+0x793)[0x1676283]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100(aggr_next_helper+0x54)[0x1677f48]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100(aggr_next+0x39)[0x1678469]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100(xchg_producer+0x1d2)[0x18e28ee]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:iix100[0x5cc140]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:/lib64/libpthread.so.0(+0x7df5)[0x7fb25c760df5]
2017-07-25 09:48:59 PID 74085:TID 140189254440704:SID 748:TXID 296535643:ERROR:/lib64/libc.so.6(clone+0x6d)[0x7fb25c2861ad]
~

Does this shed any light on the issue?

Marty

From: Dean Vernon [mailto:Dean.Vernon at actian.com]
Sent: 24 July 2017 14:39
To: Martin Bowes
Cc: Ingres and related product discussion forum
Subject: RE: [Info-ingres] Vector giving me the shits

The TMPL is the default  ... AFAIK it is a backup and is not used i.e. there is no environment variable which can be pointed to it as we can use it as an alternative.

Dean Vernon
Principal Support Engineer
Actian | Support
M +34 678 564 097
O +44 (0)1753 559552
actian.com

From: Martin Bowes [mailto:martin.bowes at ndph.ox.ac.uk]
Sent: 24 July 2017 15:06
To: Dean Vernon <Dean.Vernon at actian.com<mailto:Dean.Vernon at actian.com>>
Cc: Ingres and related product discussion forum <info-ingres at lists.planetingres.org<mailto:info-ingres at lists.planetingres.org>>
Subject: RE: [Info-ingres] Vector giving me the shits

What's the relationship between vwlog.conf and vwlog.conf_TMPL?

Marty

From: Dean Vernon [mailto:Dean.Vernon at actian.com]
Sent: 24 July 2017 14:04
To: Martin Bowes
Subject: RE: [Info-ingres] Vector giving me the shits

Default=warn is the usual setting , I always have set to info.

Dean Vernon
Principal Support Engineer
Actian | Support
M +34 678 564 097
O +44 (0)1753 559552
actian.com

From: Martin Bowes [mailto:martin.bowes at ndph.ox.ac.uk]
Sent: 24 July 2017 14:45
To: Dean Vernon <Dean.Vernon at actian.com<mailto:Dean.Vernon at actian.com>>
Subject: RE: [Info-ingres] Vector giving me the shits

I did not know that. That may be very useful.

Thanks Dean.

Marty

From: Dean Vernon [mailto:Dean.Vernon at actian.com]
Sent: 24 July 2017 13:24
To: Martin Bowes
Subject: RE: [Info-ingres] Vector giving me the shits

Sure you know but ...

In vwlog.conf

default = info:file

Call vectorwise (vwlog_reload) \g

Dean Vernon
Principal Support Engineer
Actian | Support
M +34 678 564 097
O +44 (0)1753 559552
actian.com

From: info-ingres-bounces at lists.planetingres.org<mailto:info-ingres-bounces at lists.planetingres.org> [mailto:info-ingres-bounces at lists.planetingres.org] On Behalf Of Martin Bowes
Sent: 24 July 2017 13:50
To: info-ingres at lists.planetingres.org<mailto:info-ingres at lists.planetingres.org>
Subject: [Info-ingres] Vector giving me the shits

Hi All,

VW 5.0.0 (a64.lnx/402) + p40205

Having massive problems with Vector giving up with error:
E_VW1036 Vector database is temporarily unavailable.
    Please check Vector logs (such as vectorwise.log) for more information.

The vectorwise.log shows:
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:X100 Server version x100-gcc-OPTXPROF-64-r12350 (branches/vw5.0) built on Sep 16 2016 15:27:40 with GCC 4.9.1
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:Loaded 'libhugetlbfs.so'.
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:Loaded 'libnuma.so.1'.
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:Not loaded 'libMapRClient.so': libMapRClient.so: cannot open shared object file: No such file or directory
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:Started for database 'ace_report_live' in '/dbdata1/II/ingres/data/vectorwise' with vectorsize = 1024
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:Setting the query memory limit to 734003200000 bytes
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:rank = 0, num_nodes = 1
2017-07-24 12:45:40 PID 73520:TID 140427600516992: INFO:SYSTEM:Vector nodes: (0, vw2.ndph.ox.ac.uk)
2017-07-24 12:45:47 PID 73520:TID 140427600516992: INFO:CBM:columnspace_create: nrblks 128
2017-07-24 12:45:47 PID 73520:TID 140427600516992: INFO:CBM:columnspace_create: nrblks 128
2017-07-24 12:45:47 PID 73520:TID 140427600516992: INFO:CBM:columnspace_create: nrblks 128
2017-07-24 12:45:55 PID 73520:TID 140427600516992: INFO:CBM:Setting the bufferpool limit to 202831298560 bytes
2017-07-24 12:45:55 PID 73520:TID 140427600516992: INFO:CBM:iobp_create: blksz=524288 nrblks=386870
2017-07-24 12:45:55 PID 73520:TID 140427600516992: INFO:CBM:NUMA interleave bufferpool memory.
2017-07-24 12:45:55 PID 73520:TID 140427600516992: INFO:X100SERVER:Log levels: default=warn,SYSTEM=info,SYSCALL=info,CBM=info,QUERYERROR=warn.

I've removed the parallel sorting.

I've ensure stacksize is at 8M.

Anyone got any other ideas?

Martin Bowes
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.planetingres.org/pipermail/info-ingres/attachments/20170725/d6fe2808/attachment.html>