[OPEN-ILS-DEV] Frequent error message from Evergreen 2.2

[Robert Heller]

Latest bit of news:

Its been about a week since I changed the version of xulrunner being
used with the Evergreen 2.3 staff client. We are now running the
Evergreen 2.3 staff client (downloaded from the C/W Mars site) with the
stock 32-bit version of xulrunner *natively* (NOT using WINE) under
CentOS 5.9 (Linux). This is version 10.0.12 of xulrunner. We are
running this on both workstations (circulation desk and back office
[cataloging mostly]). These workstations are Pentium 4's with 2gig of
RAM.  Our connection to C/W Mars is via a full T1 line (1.5mbit).  The
catalogger has reported 0 (*ZERO*) incidents of the (infamous?) "Error
in volume_copy_creator.js, g.doc_id is not valid" message.  I don't
know if this is luck or what.

I have also noticed that the circulation desk system is running better
and my logging of memory use suggests that the memory leak tapers off
(VIRT memory reaches a plateau). Things still slow down by the
afternoon or early evening and restarting the Evergreen 2.3 staff
client helps.

Our Linux boxes have *never* actually crashed -- Linux deals with
excess load or memory starvation much more gracefully than MS-Windows
-- Linux might get slow under excessive load, but otherwise keeps
chugging along (and recovers once the excessive load is removed -- eg
when the overextended process is killed or exits).


At Fri, 28 Sep 2012 13:47:02 -0400 Robert Heller <heller at deepsoft.com> wrote:

> 
> At Fri, 28 Sep 2012 16:12:30 +0000 Alex Reczkowski <areczkowski at cwmars.org> wrote:
> 
> > 
> > Dear Robert & Jason,
> > Just weighing in from Pittsfield (also in C/W MARS) -- we were
> > plagued with this error but found that since we upgraded our internet
> > line (from 7Mbps/2Mbps to 35Mbps/5Mbps), it has become quite rare.  So,
> > speed and reliability of the connection seems to have taken care of the
> > issue.
> > Now that I've typed this, I figure that I probably haven't added much
> > to your knowledge, but if you can think of anything I can check or test
> > here (since we had the problem and seem to have beaten it), please let
> > me know.
> 
> Looking at the bug report, it appears that the problem is a 'race
> condition'.  Race condition bugs are generally timing related (eg one
> process gets ahead of another and thing go south when syncronization is
> lost).  Anything that makes the timing worse tends to make the race
> condition worse or more noticable (occuring more frequently).  This
> very much seems to be what is going on.
> 
> It also appears that a bug fix is in progress...
> 
> > Sincerely,
> > Alex
> > 
> > Alex Reczkowski
> > Supervisor, Technical Services
> > Berkshire Athenaeum, Pittsfield's Public Library
> > 1 Wendell Avenue
> > Pittsfield, MA 01201
> > 413-499-9480 ext. 115
> > fax: 413-499-9489
> > 
> > 
> > 
> > ----------------------------------------------------------------------
> > 
> > Message: 1
> > Date: Thu, 27 Sep 2012 12:40:23 -0400
> > From: Robert Heller <heller at deepsoft.com>
> > Subject: [OPEN-ILS-DEV] Frequent error message from Evergreen 2.2
> > 	Staff	Client: "Error in volume_copy_creator.js, g.doc_id is not valid"
> > To: open ils dev <open-ils-dev at list.georgialibraries.org>
> > Cc: Tim Spindler <tspindler at cwmars.org>
> > Message-ID: <201209271640.q8RGeN9a008060 at sharky2.deepsoft.com>
> > 
> > I am the IT guy for the Wendell Free Library, a member of the C/W Mars library system here in Western Mass.  We are getting an error message when the librarians try to catalog new books into the library system's database. The error message is:
> > 
> > "Error in volume_copy_creator.js, g.doc_id is not valid"
> > 
> > And it comes up when the cataloger tries to load the record for a book to edit it (fix incorrect fields, etc.).
> > 
> > We are getting this error message way too frequently, but most of the other libraries in our system are not getting this error more than very rarely.  We are *unique* in that we are using Linux machines to run the staff client.  We are also one of two libraries that are using a T1 line as our network connection to the library system server (the other library also on a T1 line is also having this problem as well).
> > 
> > Our machines are small form factor diskless Pentium 4s with 2Gig of RAM.  They boot over the network and NFS mount their file systems. We are running CentOS 5.8 and are using a repackaged version of XULRunner from CentOS 5.7 (xulrunner19-1.9.2.26-1) to run the Evergreen 2.2 Staff Client.  We are NOT using wine to run the MS-Windows version.
> > 
> > Is this a known problem?  Is there some workaround?  Is there something we can do to help fix this problem.  The error occurs as part of the cataloging process and has pretty much brought our cataloging process to a standstill.
> > 
> 

-- 
Robert Heller             -- 978-544-6933 / heller at deepsoft.com
Deepwoods Software        -- http://www.deepsoft.com/
()  ascii ribbon campaign -- against html e-mail
/\  www.asciiribbon.org   -- against proprietary attachments