FYI, you're not alone -- several people have reported=20
this problem in upgrading.
Starting with the next release, this information will be
stored in the database rather than the filesystem. This
should help to make the migration process go more smoothly.
From: Ronda A. Grizzle [mailto:rag9b@...]
Sent: Tuesday, September 30, 2003 3:02 PM
To: Michael J. Giarlo; fedora-users@...
Subject: Re: [Fedora-users] OBJID generation resets after upgrades
I have this information from Ross about retaining current pid numbering=20
when upgrading your Fedora repository:
--begin pasted text--
PID generation is controlled by special PID log files located in:
where YYYY is 4-digit year, MM is 2-digit month, and DD is 2-digit day
When switching versions of repositories and you want to resume the PID=20
numbering where you left off in the initial repository, you need to do =
1) Remove any logs that may have been accidentally created in the new=20
version. PID files are sorted by date so you want to remove any errant=20
ones that will have a date *later* than the last log in the previous=20
2) Copy all of the PIDGeneration logs from the old installation to the =
one, then start the new server.
Note: Be sure both repositories have the same PID namespace defined in=20
their fedora.fcfg files.
e.g., on unix
cp $FEDORA_HOME_OLD/server/pidgen/*.log =
where OLD/NEW represent the install directories for old/new versions
--end pasted text--
If you have any questions about this, please do let me know.
--On Tuesday, 30 September, 2003 10:24 AM -0400 "Michael J. Giarlo"=20
> After running the 1.1 and 1.1.1 upgrades to Fedora, we noticed a =
> odd behavior. When ingesting objects without a specified OBJID,
> auto-numbering seems to reset to 1 within our pidNamespace. I believe
> we were up above 100 in the auto-numbering scheme in version 1.0 when =
> upgraded to 1.1, and perhaps a bit higher after dealing with this
> problem just to see it crop up again after the 1.1.1 upgrade. Now =
> almost all of the OBJIDs below 100 are taken within our pidNamespace,
> ingesting an object with an unspecified OBJID can be a time-consuming
> and tedious task, as Fedora 1.1.1 increments its internal counter by 1
> on each ingest. As you can imagine, needing to ingest an object =
> of times before Fedora finds an unused OBJID is quite bothersome. :)
> This leads me to wonder where Fedora stores the highest auto-generated
> OBJID -- in the FedoraObjects database, or in a file somewhere? -- or =
> the value is not stored at all, but generated somehow, e.g. out of the
> pidgeneration logfiles. With this knowledge, I can just change its =
> to our highest used OBJID, and our problem goes away.
> Has anyone else seen this behavior? Can anyone recommend a workaround
> or solution?
> Additionally, is this the desired behavior? I.e. will this happen =
> time we upgrade?
> Thanks in advance!
> Michael J. Giarlo Scholarly Communication =
> Network/Systems Administrator Rutgers University Libraries
> --Email: sccadmin@... Cell: +1 732-310-0976
> --Phone: +1 732-932-8573 x182 Fax: +1 732-932-1386
> Fedora-users mailing list
Ronda A. Grizzle
Technical Coordinator, Fedora Project
Digital Library Research & Development
Fedora-users mailing list