MySQL Bugs: #27372: agent shows: job_collect_mysql.c.361: executing the regex for .. (78) failed: -1

Bug #27372	agent shows: job_collect_mysql.c.361: executing the regex for .. (78) failed: -1
Submitted:	22 Mar 2007 13:38	Modified:	22 May 2009 8:26
Reporter:	Carsten Segieth	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Enterprise Monitor: Agent	Severity:	S3 (Non-critical)
Version:	1.1.0.4899, 2.0.0.7092	OS:	Windows (Windows 2003)
Assigned to:	Jan Kneschke	CPU Architecture:	Any
Tags:	mem_20_maint, mer 111, mer 120

Description:
Agent log shows tons of messages like

2007-03-22 14:16:00: (critical) c:\cygwin\home\mysqldev\merlin\src\mysql-service-agent-1.1.0.4899r4901\agent\src\job_collect_mysql.c.361: executing the regex for innodb pattern "\d+.\d* inserts\/s, \d+.\d* updates\/s, \d+.\d* deletes\/s, (\d+.\d*) reads\/s" (78) failed: -1

How to repeat:
- note sure I'm able to repeat it
- agent running since > 2 weeks against a 1.1.0.4876 Merlin (net-qa2)
- since yesterday (when I used the Graphs page) these messages occur in the log

I am seeing this as well. Note - I have set the purge for logs and data to 1 week, and the time when these entries started coincides with the time the dashboard was started (different date).

I was getting the same errors too on only one of my servers and my innodb buffer pool graph was blank.  They stopped right after I issued a "show innodb status" on that server.

fixed in [5646].

SHOW INNODB STATUS might contain too many deadlocks, lock-records, transactions at the top of the output. As the whole output is limited to 64kb.

Still happens with 1.2.0.6610, but *only* against a 4.0.30 server. Same check with a mysqld version 4.1, 5.0, 5.1 and 6.0 didn't show the message in the log. From this I reduced the S/P setting.

2007-07-13 17:25:09: (message) mysql-query: SHOW /*!40100 ENGINE */ INNODB STATUS
2007-07-13 17:25:09: (critical) agent/src/job_collect_mysql.c.1145: executing the regex for innodb pattern "History list length (\d+)" (25) failed: -1
2007-07-13 17:25:09: (message) mysql::innodbstatus->innodb_bp_modified_pages = 0

full log is available in qauser@net-qa1:/nfstmp1/merlin/agent/1.2.0.6409/rhas4-x86/net-qa1/log/10.100.1.224.net-qa2.3340.log

fix committed in [6970] to trunk/

r6970 trunk

Added to the changelog for version 1.2.

Re-opening this bug.

This is repeatable using the latest beta version, 1.2.0.7481 with 5.0.x.

However, what I have found it that it is related to bug 19825:

http://bugs.mysql.com/bug.php?id=19825

This is a bug where some of the SHOW INNODB STATUS is truncated.  So, until bug 19825 gets fixed, this one is not truly fixed.

Per Leith: 

(1) for 2.0 maint release, lower the log level (make it a warning)
(2) then regard this as a feature request to read innodb status from status file if available

committed and push into trunk and rel-2.0:

revno: 1215
committer: jan@mysql.com
branch nick: trunk
timestamp: Tue 2009-03-03 00:13:44 +0100
message:
  hide PCRE_ERROR_MATCH in case we have SHOW INNODB STATUS-buffers of 60k or more (fixes #27372)

revno: 1224
committer: jan@mysql.com
branch nick: rel-2.0
timestamp: Tue 2009-03-03 00:16:21 +0100
message:
  hide PCRE_ERROR_MATCH in case we have SHOW INNODB STATUS-buffers of 60k or more (fixes #27372)

Patch installed in versions => 2.0.5.7150.

Changed status by mistake. 7150 was only a server build.

Patch installed in versions => 2.0.5.7150.

tested OK with 2.0.5.7153

An entry has been added to the 2.0.5 changelog:

Multiple errors showed in the agent log after issuing a SHOW INNODB STATUS statement. The InnoDB Buffer Pool graph also went blank.