zenoss / zenpacks.zenoss.postgresql Goto Github PK

I'm seeing alerts generated for

postgres failure: [Errno 2] No such file or directory: '/tmp/.zenoss_postgres_5c45ef624e0607d6d431b8138f87bd68'

I'm presuming this tmp file is created by the zenpack itself, so not sure why this would be occuring.

Hi Chet.

We've recently upgraded our Zenoss Core from 3.2.0 to 3.2.1 and seems that somehow it did break our PostgreSQL plugin. This is the output from zentestcommand:

Traceback (most recent call last):

File "/usr/local/zenoss/zenoss/libexec/poll_postgres.py", line 23, in

from util import PgHelper

ImportError
:
No module named util

2012-05-16 15:29:57.745 EDT pjpn [unknown] postgres 192.136.136.181 idle in transaction 08P01 pjpn: LOG: unexpected EOF on client connection
2012-05-16 15:29:57.745 EDT postgres [unknown] postgres 192.136.136.181 idle in transaction 08P01 postgres: LOG: unexpected EOF on client connection

Looks like its not ending the connection gracefully

Similar to issue #6 by magnusp, we have many database servers with multiple databases. in turn with multiple schemas.

When browsing the PostgreSQL Tables component for the device, there are just too many tables in there.

I'd like to request that the database/tables/(schemas) be organized in 'folders' so that you can just view the tables within a specific database/tables/(schemas) combo. This would probably also help with report generation (To compare master and replica server performance)

Just looking at the layout of the files on the filesystem, and even having the components laid out in the same way would be just great

This zenpack is not closing connections. My logs are full of these.
Running 1.0.8 on PGSQL 9.3

2015-07-29 15:46:32.202 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""
2015-07-29 15:46:32.211 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""
2015-07-29 15:46:32.211 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""
2015-07-29 15:46:32.211 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""
2015-07-29 15:51:32.192 ,"too many connections for role ""service_zenoss""",,,,,,,,,""
2015-07-29 15:51:32.200 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""
2015-07-29 15:51:32.200 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""
2015-07-29 15:51:32.200 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""
2015-07-29 15:51:32.201 ,"unexpected EOF on client connection with an open transaction",,,,,,,,,""

Hi folks, we've got Zenoss Core 3.2.0 and Postgres zenpack working OK, but we frequently get this kind of event alerts:

Device: db-master
Component:
Severity: Error
Time: 2011/11/10 23:10:13.000
Message:
postgres failure: Unterminated string starting at: line 1 column 8177 (char 8177)

Clear message comes always 5 minutes later:

Event: 'postgres failure: Unterminated string starting at: line 1 column 8177 (char 8177)'
Cleared by: 'postgres connectivity restored'

After Zenpack install, we've started to get many of those messages into postgres log file:
LOG: unexpected EOF on client connection

Googling around seems to be a client problem:

http://postgresql.1045698.n5.nabble.com/LOG-unexpected-EOF-on-client-connection-td3204658.html

One suggestion is to tweak tcp_keepalive settings for the client.

Alexander Fortin [email protected] requested support for monitoring over SSH to the PostgreSQL server instead of the native PostgreSQL protocol. This would be better for situations where you don't want to open the PostgreSQL server to remote connections.

Data sources work in zenoss 5 but components don't populate. i.e. you get the overall server stats, but the individual databases and tables that once appeared under device components is missing. I am having similar issues with other zenpacks.

In my environment we have 18 postgres servers with ~25 databases each, containing ~4000 schemas per database. I've only just started using the Zenpack, but from what I understand each table, per schema, is monitored/gathered statistics from. For me, that would mean in-the-millions RRDs and an unacceptable load on the servers for each poll.

I propose two mutually exclusive settings that either includes or excludes schemas from being monitored. This could be implemented as (a list of?) regular expressions against the schemaname and/or relname columns of pg_stat_user_tables.

Hi,

When I tried to use ZenPacks.zenoss.PostgreSQL on a quite new/empty database I've got "postgres failure: float division". I suspect there is no check for 0 on some value. I modified poll_postgres.py to get some stact trace and when running via "Test Against Device" I've got:

Preparing Command...
Executing command $ZENHOME/libexec/poll_postgres.py <_> server against <_>
File "/opt/zenoss/libexec/poll_postgres.py", line 187, in printJSON
data = self.getData()
File "/opt/zenoss/libexec/poll_postgres.py", line 125, in getData
databases = pg.getDatabaseStats()
File "/opt/zenoss/ZenPacks/ZenPacks.zenoss.PostgreSQL-1.0.3-py2.6.egg/ZenPacks/zenoss/PostgreSQL/util.py", line 172, in getDatabaseStats
xactRollbackPct = (float(row[4]) / xactTotal) * 100
*** float division *** print_tb: {"events": [{"eventClassKey": "postgresFailure", "summary": "postgres failure: float division", "severity": 4, "eventKey": "postgresFailure"}]}
DONE in 1 seconds

ZenPacks.zenoss.PostgreSQL 1.0.3 (for Python 2.6)
zenoss-3.2.0-1258.el5
PostgreSQL 9.1

Regards
Marcin