[Postgres-xl-general] Failed to get pooled connection

Andrei Martsinchyk andrei.martsinchyk at gmail.com
Fri Sep 19 00:02:29 PDT 2014


To execute complex query one datanode may establish multiple connections to
another one.
To figure out for particular query you can look at the query plan.
Find rows looking like

   ->  Remote Subquery Scan on all (d1,d2)  (cost=1150927.67..5161944.87
rows=56323319 width=0)

Each such row assumes making new connections to the listed datanodes.
Keep in mind, there could be multiple instances of Remote Subquery Scan
running, if there is another Remote Subquery Scan above, which makes
multiple connections.


2014-09-19 9:40 GMT+03:00 Aaron Jackson <ajackson at revionics.com>:

>      Didn't see any networking issues.  And there's no cluster
> reconfiguration, it happened today when I happened to bring up a new client.
>
>  Andrei, good point, I didn't even think about the fact that we have
> datanode to datanode connections here.  Coming from PGXC, we'd multiple
> directly as a factor of the coordinators.  I'll try bumping the
> connections, but I don't think we've been even close to the connection
> limit.  But you never know where resources could be getting scoured from in
> a datanode to datanode scenario.  Good call.
>
>  *Aaron*
>      ------------------------------
> *From:* Pavan Deolasee [pavan.deolasee at gmail.com]
> *Sent:* Friday, September 19, 2014 12:47 AM
> *To:* Aaron Jackson
> *Cc:* postgres-xl-general at lists.sourceforge.net
> *Subject:* Re: [Postgres-xl-general] Failed to get pooled connection
>
>
>
> On Fri, Sep 19, 2014 at 3:19 AM, Aaron Jackson <ajackson at revionics.com>
> wrote:
>
>>      Occassionally, my servers will fail with the daunting "Failed to
>> get pooled connection" error message.  Usually, I will just restart (kick)
>> the datanodes and coordinators to resolve the issue.  But there's no
>> indicator about why it happens and it's frequency is so inconsistent that
>> its just a little unnerving to see it when it happens.
>>
>>  Any thoughts about the root cause of this?  Anything I can look into in
>> detail?
>>
>>
>  Do you see any networking issues here? But if restarting
> datanodes/coordinators resolve the error, then it looks unlikely.
>
>  Does the cluster undergo any reconfiguration?
>
> Next time this happens, it might be worth trying to connect the datanode
> directly via psql (preferably from the coordinator node) and see if that
> works ok. I wonder if you might be hitting some limits such as open file
> descriptors on the datanode.
>
>  Thanks,
> Pavan
> --
> Pavan Deolasee
> http://www.linkedin.com/in/pavandeolasee
>
>
> ------------------------------------------------------------------------------
> Slashdot TV.  Video for Nerds.  Stuff that Matters.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=160591471&iu=/4140/ostg.clktrk
> _______________________________________________
> Postgres-xl-general mailing list
> Postgres-xl-general at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/postgres-xl-general
>
>


-- 
Andrei Martsinchyk

StormDB - http://www.stormdb.com
The Database Cloud
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.postgres-xl.org/private.cgi/postgres-xl-general-postgres-xl.org/attachments/20140919/2623e61e/attachment.htm>


More information about the postgres-xl-general mailing list