[
https://issues.apache.org/jira/browse/TS-758?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14492139#comment-14492139
]
Greg O'Rawe commented on TS-758:
--------------------------------
I am seeing this problem on a customer deployment, and the hostdb allocates
memory for DNS entries which return multiple IP addresses as described.
However, if an entry is added to the hostdb and a request is made to that
domain after the TTL has expired, heap memory is allocated again for the same
domain which already exists in the table. The entry itself is removed and
re-inserted, but there is a further immediate allocation for any round-robin
data returned. The memory steadily grows (e.g. by 160 bytes each time). This
looks like a bug. It is version 3.2.0 being used here, but I don't think the
code has changed in later versions. If the entry already exists in the hostdb
but the TTL has expired, then surely there should not be another allocation for
the same round robin entries after the DNS lookup?
Greg
> HostDB fails to insert RR entries (allocation fails) claiming DB is full
> ------------------------------------------------------------------------
>
> Key: TS-758
> URL: https://issues.apache.org/jira/browse/TS-758
> Project: Traffic Server
> Issue Type: Bug
> Components: DNS
> Affects Versions: 2.1.8
> Reporter: Leif Hedstrom
>
> After running in forward proxy for a few days, I got a bunch of these errors:
> [May 3 17:13:11.574] Server {140488838387680} WARNING: out of room in hostdb
> for round-robin DNS data
> [May 3 17:13:11.726] Server {140488838387680} WARNING: out of room in hostdb
> for round-robin DNS data
> [May 3 17:13:11.736] Server {140488838387680} WARNING: out of room in hostdb
> for round-robin DNS data
> [May 3 17:13:14.072] Server {140488821982992} WARNING: out of room in hostdb
> for round-robin DNS data
> [May 3 17:13:17.114] Server {140488838387680} WARNING: out of room in hostdb
> for round-robin DNS data
> I'm not sure why it thinks it's out of room, there is no way I've filled my
> hostdb cache (I'm fairly certain). I'm not able to reproduce the problem
> (easily).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
