[ClusterLabs] 答复: No slave is promoted to be master

范国腾 fanguoteng at highgo.com
Thu Apr 12 04:31:39 EDT 2018


Thank you very much for help check this issue. The information is in the attachment. 

I have restarted the cluster after I send my first email. Not sure if it affects the checking of "the result of "crm_simulate -sL"

We started the performance test at about 2018-04-11 17:40. 

The problem started from the following log (var/log/messages in node1):
Apr 12 02:28:12 sds1 crmd[2934]:  notice: db1-pgsqld_monitor_15000:954 [ /tmp:5866 - no response\n ]

The postgres log show:
2018-04-12 02:28:12.904 CST,"postgres"," postgres ",1175,"[local]",5ace53b9.497,1,"idle",2018-04-12 02:28:09 CST,303/0,0,LOG,08006,"could not send data to client: Broken pipe",,,,,,,,,"pg_isready"
2018-04-12 02:28:12.904 CST," postgres "," postgres ",1175,"[local]",5ace53b9.497,2,"idle",2018-04-12 02:28:09 CST,303/0,0,FATAL,08006,"connection to client lost",,,,,,,,,"pg_isready"
2018-04-12 02:28:14.797 CST,,,17030,,5acdd8ad.4286,3,,2018-04-11 17:43:09 CST,,0,LOG,00000,"received fast shutdown request",,,,,,,,,""

Then the master is restarted and it could not start(that is ok and we know the reason)。Please help check why does pacemaker not promote one of the slave?

发件人: Jehan-Guillaume de Rorthais [mailto:jgdr at dalibo.com] 
发送时间: 2018年4月12日 15:57
收件人: 范国腾 <fanguoteng at highgo.com>
抄送: Cluster Labs - All topics related to open-source clustering welcomed <users at clusterlabs.org>
主题: Re: [ClusterLabs] No slave is promoted to be master

On Thu, 12 Apr 2018 03:14:52 +0000
范国腾 <fanguoteng at highgo.com> wrote:
> We have three nodes in the cluster. When the master postgres resource in one
> node(db1) crashed and could not start any more, we hope that one of the slave
> node(db2,db3) could be promoted be master. But it does not.
> [cid:image001.jpg at 01D3D24F.03F259A0]
> Here is the log
> db1:
> [cid:image002.jpg at 01D3D24F.03F259A0]
> Db2:
> [cid:image007.png at 01D3D24E.200050D0]
> Db3:
> [cid:image008.png at 01D3D24E.200050D0]

Could you please provide:

* your full logs from all nodes as textual (compressed) files?
* the full setup of the cluster
* the result of "crm_simulate -sL"

-------------- next part --------------
A non-text attachment was scrubbed...
Name: info.tar.gz
Type: application/x-gzip
Size: 1192602 bytes
Desc: info.tar.gz
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20180412/a2ce7c45/attachment-0002.bin>

More information about the Users mailing list