monit start all fails on 5.19.0

[Former user]

Hi,

Upgraded from 5.14 to 5.19.0 and found a regression with our monit configuration. Specifically, "monit start all" and "monit stop all" now fail without starting or stopping any services. For example:

$ monit start all
There is no service named "FS="

$ monit stop all
There is no service named "FS="

Note that we have several service checks of type filesystem defined as follows:

check filesystem FS=/foo/bar with path /foo/bar

It seems that 5.14 monit would correctly handle <service name> of this form but in 5.19 this name causes a regression.

We would really really like to not change the service names to adopt or trip over this issue in some other manner.

Thanks for your consideration. -sandy

[Sandy C]

Sorry - unable to correct type-o's and atlassian induced formatting errors in the above. Basically "monit stop all" returns the string 'There is no service named "FS="'.

If all the services that begin with the string "FS=<some real absolute path>" are removed, then monit 5.19.0 seems to work. (Historically, the service checks are generated at orchestration time, and as such all the filesystem and directory checks are generated with the prefix "FS=" and "DIR=" pre-pended to the actual path.)

As such, we are hoping to not change all that, if that is the issue, so to adopt 5.19. Thanks!

[Tildeslash]

Fixed: Issue #486: Fix handling of actions for services with [?&=/] characters in their name.

→ <<cset afbbbe3f0fec>>