[OPEN-ILS-DEV] router@localhost/open-ils.storage IS NOT CONNECTED TO THE NETWORK!!!

[Dan Scott]

On 04/04/07, Harry Bochner <Harry.Bochner at biogenidec.com> wrote:
>
> Dan Scott wrote:
> > Have you tried checking the list archives from about March 22nd? Cory
>  > was experiencing a similar problem (around here:
>  >
> http://list.georgialibraries.org/pipermail/open-ils-dev/2007-March/000484.html)
>
>
> I've read through that thread, without any clear conclusion.
> In his case opensrf.setting, iirc, wasn't starting. That's
> starting OK for me.
> And the answer in his case was the transport/username setting marked
> "do not change this".
>
> Well, what I have there is "router", r.t. "mylogin", but _I_ haven't
> changed it, that's what I found in the example files for 1.0.3.2. And
> as far as I understand the config files, the name "router" is being used
> consistently. But, just to be sure, I tried changing the relevant
> occurrences to "mylogin", the way they were in the earlier thread: no
> change in the symptoms.
>
> So I don't that's it.
>
> Earlier in that thread, there was a suggestion to start new logs, and grep
> for ERR or warning. All I find is
> router.log:router 2007-04-04 16:11:09
> [WARN:10696:osrf_router_main.c:10:] Received signal [2],
> cleaning up...
> router.log:router 2007-04-04 16:11:09 [WARN:10696:osrf_router.c:60:] Top
> level select call failed with errno 4
>
> And I think that just has to do with osrf_ctl.sh -a stop_all
>
> > And yes, osrf_ctl.sh definitely should start all of the open-ils
>  > services for you.
>
> OK, good. If someone can give me some pointers to the code that does it,
> I can try to trace through it.  There's a _lot_ of code here ;-) and I don't
> know my way around it yet.
>
> The impression I'm getting is that my openils.xml file is not being read, or
> at least
> used. You might recall that the IDL element I had there was ignored. And
> when I
> run osrf_ctl.sh -a start_all, I can see that the other two files have been
> read,
> while openils.xml hasn't:
> $ ls -lu o*.xml
> -rw-r--r--  1 bochnerh reslib 27415 Apr  4 16:25 openils.xml
> -rw-r--r--  1 bochnerh reslib  4015 Apr  4 16:28 opensrf_core.xml
> -rw-r--r--  1 bochnerh reslib  8904 Apr  4 16:28 opensrf.xml
>
> Here's a copy of my config files, in case that helps:
>
>
>
> --
>  Harry

Hi Harry:

A couple of things... first of all, I'm working with HEAD, so your
mileage may vary.

A change in CVS on March 19th broke Evergreen for me, with symptoms
very similar to what you're seeing. The problem was that the advice to
set certain variables = to the output of `hostname -f` remained in
place, while a more portable alternative (perl -MNet::Domain) was put
in its place in the code. Unfortunately, the more portable alternative
actually returns a different value - well, it does on my machine
anyways. I get "evergreen.openils" with the former, and just
"evergreen" with the latter, which plays havoc with an opensrf.xml
file that contains a <evergreen.openils> block. And by "havoc" I mean
none of the corresponding services get started and opensrf-c
segfaults.

This problem is being tracked in bug 534
(http://www.open-ils.org/cgi-bin/bugzilla/show_bug.cgi?id=534). When I
changed <evergreen.openils>...</evergreen.openils> in opensrf.xml to
<evergreen>...</evergreen>, everything started working again. Phew!

Second, and this is just a heads-up, in the 1.1 branch the developers
opted to reduce the confusion of having bootstrap tell you to point to
opensrf.xml while your config was actually placed in openils.xml by
deprecating openils.xml and making opensrf.xml king. The Makefile
copies openils.xml.example over top of opensrf.xml.example to reflect
the additional services offered by Open-ILS, so it's all good.

Dan