This is an old issue, and we've made a number of fixes since (CWD-3769, CWD-3692). Due to the substantial changes, I'm going to mark this issue as obsolete. If we have similar cases still reproducible then they are better handled as new issues.

Hi dtheodore,

We've just released Crowd 2.7.2. It contains fixes related to the cwd_token table. Can you please confirm if the problem is still present in 2.7.2? Thank you.

Confirmed on Crowd 2.7.1 with Oracle 11.2.0.4.0 RAC

This issue was first reported against an old version of Crowd (2.0.4). The code dealing with the cwd_token table has changed substantially in newer releases. Similar problems have been reported with Crowd 2.7.0 and 2.7.1 (see CWD-3769 and related issues). The upcoming Crowd 2.7.2 release addresses these problems and we believe it may fix this issue as well.

dasha1, as discussed on our bug fixing policy, there are a variety of factors which go into influencing our priorities when determining which bugs to fix; this one is currently not on our short term roadmap.

If you require further support, please discuss on your support case. (jira.atlassian.com is for tracking bugs & feature requests, but is not well suited to 1 on 1 support and troubleshooting; public comments mean that sensitive information may be accidentally divulged.)

Hi Team,

We have configured JIRA, confluence,bamboo, fisheye/crucible with crowd for authorization and authentication. Also crowd is configured with 4 directories for 4 LDAPS. We receive the below errors quite frequently and users are not able to login.

2014-04-14 17:17:13,911 http-443-40 ERROR [crowd.manager.application.ApplicationServiceGeneric] Directory 'ASIA LDAP' is not functional during authentication of 'fit-build'. Skipped.
2014-04-14 17:18:27,333 http-443-3 ERROR [crowd.manager.application.ApplicationServiceGeneric] Directory 'GLOBAL LDAP' is not functional during authentication of 'redkarv1'. Skipped.

I was suggested by your analyst " Septa Cahyadiputra [Atlassian]" that my issue is related to this bug.
Also I have changed the "Authentication Token Storage" to " Memory Cache" for one of the directories. However, I do not see any improvement.

Kindly assist on this ASAP.

Regards,
Aurobindo.

You should fix this issue or at least recommend to use in-memory tokens in the documentation (https://confluence.atlassian.com/display/CROWD/Oracle).

Please note that this issue also happens with Postgres, not only with Oracle.

We would prefer to continue to use database-based tokens and believe Atlassian should fix this issue. From my point of view, issues related to logged exceptions cannot be rated with minor/low priority!

We do not have a high load on Crowd. Both memory as well as CPU activity look very good.
And: the only downsize you mentioned is the reason for prefering to store the tokens into the database.