OAI resume link returns nothing (resumptionToken issue?)

Software Support
,
1

Hi fellows,

Recently an indexer reported to us that our OAI “resume” link was not working fine.
We are on OJS 2.4.5, and we didn’t noticed before this issue before because our “oai_max_records” is set to 1000 registers.

You can see the problem here:
http://papers.uab.cat/oai?verb=ListRecords&metadataPrefix=oai_dc

When you click in the “resume” link at the very bottom of this list (under “There are more results.” label) you got a “ERR_EMPTY_RESPONSE” error.

If I extend “oai_max_records” to 1500, a php time out jumps… And yes, I can change max_execution_time in my server, but journals are growing so this will be a temporary solution.

Any clue about how to fix the “resume” feature?

Thanks for your help,
m.

2

I checked this functionality in OJS 2.4.8 for one of our larger journals. It works as expected.

Do you have anything in the error log that might help?

3

A harvaster is pushing me so as a temporary solution I changed the max_execution_time.

Only this record looks related:
[08-Jul-2016 12:49:41 Europe/Berlin] NOTICE: Use of undefined constant Registrarse - assumed 'Registrarse' (/home/ojs/source/versions/papers/plugins/importexport/datacite/classes/DOIExportPlugin.inc.php(241) : assert code:1)

At first glance I though was missing chain in the lang files… but is not.

4

That particular notice seems unlikely to be the culprit. I don’t see anything recent in bugzilla or Git Issues about the OAI resumption token for 2.4.5 to 2.4.8, but I wonder if there is something in the version difference. Any change you could spin up a dev site with 2.4.8 and the data from papers.uab.cat?

6

I found a segfault (11) in my apache’s log:

[Thu Jul 28 15:44:33 2016] [notice] child pid 13412 exit signal Segmentation fault (11)

I suspect a clean 2.4.8 OJS will work like a charm, but then the issue will still be there.

Apache’s log is not very verbose, but it’s an starting point. :wink:
If I get stacked I will try with a clean installation.

Cheers,
m.

7

Hi @marc,

Hmm, that sounds like a buggy PHP. Has anything changed there?

Regards,
Alec Smecher
Public Knowledge Project Team

8

Not recently, but the problem raised recently so it could be there since the beginning of time … :blush:

Any suggestion about how to debug this without gdb and it’s abusive execution stacks?

Changing apache log to “debug” don’t help much…

Cheers,
m.

9

Hi @marc,

Hmm, unfortunately I can’t help much with that – but if you’re getting a reliable segmentation fault message, then something is definitely wrong beyond the PHP code’s responsibility. It shouldn’t be possible to trigger a segfault from within PHP. Checking for an easy PHP update would be a likely place to start.

Regards,
Alec Smecher
Public Knowledge Project Team

1 Like