[Postgres-xl-general] pgxc_ctl failing

Pavan Deolasee pavan.deolasee at gmail.com
Sun Oct 26 23:07:29 PDT 2014


On Thu, Oct 23, 2014 at 11:36 PM, John Melesky <john.melesky at rentrakmail.com
> wrote:

> After further experimentation, it appears that 'pgxc_ctl prepare' will, if
> given a valid conf file, overwrite out a default conf file instead. And, if
> given no pgxc_ctl conf file, or an invalid one, dumps core.
>
> This suggests to me that my best bet is to skip the 'prepare' step, and go
> directly to implementing a conf file and issuing 'init all'.
>
> Does that sound accurate?
>

Yeah, I usually do that. But the prepare step generates a good template to
build on. I'm kind of surprised why did it generate an invalid conf file in
the first place though. I tried on my setup and your original steps work OK
for me. The conf file has the gtmSlaveName (which is a new addition since
the last release was made).

Do you have some more information regarding this? Had you had generated the
conf file with old pgxc_ctl i.e. the one from the released sources?

The segmentation fault is unfortunate and should be handled cleanly. I'll
look into it.


> I admit that I'm confused as to the usage and purpose of this utility (but
> not to its necessity -- as mentioned, manually configuring a cluster was
> just as fruitless and more frustrating).
>

Right. We are discouraging people to manually setup the cluster and instead
use pgxc_ctl. Its also slightly rough around the edges, but has been
improved a lot since the last beta release. We intend to make it even
better and easier to use.

Thanks,
Pavan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.postgres-xl.org/pipermail/postgres-xl-general-postgres-xl.org/attachments/20141027/f18f6f34/attachment.htm>


More information about the postgres-xl-general mailing list