0001174: mod_caucho causes apache failure when srun host is down

Viewing Issue Simple Details [ Jump to Notes ]

[ View Advanced ] [ Issue History ] [ Print ]

Category

Severity

Reproducibility

Date Submitted

Last Update

0001174

[Resin]

major

always

06-05-06 14:27

06-28-06 09:54

Reporter

jdane

View Status

public

Assigned To

ferg

Priority

high

Resolution

fixed

Status

closed

Product Version

3.0.19

Summary

0001174: mod_caucho causes apache failure when srun host is down

Description

If when apache starts the host pointed to by ResinConfigServer is down or otherwise unreachable, apache will be unable to serve any content at all, even static content with no connection to resin.

the situation in which this can happen is rare enough, but it came as a complete surprise to me when it did happen that a failed resin could take out apache completely.

it seems that connections to the ResinConfigServer can have a timeout applied, but that such a timeout is not applied on the initial requests. that is, I observed cse_connect_wait being called, rather than cse_connect.

this is admittedly a tough nut, but MO is that the current behavior is unacceptable. perhaps the initial request could have a connect timeout applied, and further that a flag could be checked such that no attempt will be made to connect to the srun for the next N minutes.

Additional Information

Attached Files

Relationships

Notes
(0001265) jdane 06-05-06 14:34	I should mention the workaround I came up with: put the ResinConfigServer directive within <Location> blocks. this works, but it seems to be (?) defeating the purpose of the directive.

Issue History
Date Modified	Username	Field	Change
06-05-06 14:27	jdane	New Issue
06-05-06 14:34	jdane	Note Added: 0001265
06-28-06 09:54	ferg	Assigned To	=> ferg
06-28-06 09:54	ferg	Status	new => closed
06-28-06 09:54	ferg	Resolution	open => fixed
06-28-06 09:54	ferg	Fixed in Version	=> 3.0.20

Mantis 1.0.0rc3[^]

29 total queries executed.
26 unique queries executed.