Handling out of disk space

since JEMH 3.3.12

Prior to 3.3.12, if your Jira ran out of disk space and nobody noticed, JEMH would hit a variety of failures as a result, possibly including failure to create an audit event for an email, resulting in the failure to process it, and failure to forward the incoming mail as an attachment, potentially loosing data.

Process

JEMH will evaluate the disk space prior to processing every email, a lockfile is used to flag whether we’ve already issued ‘forward’ notification {{JIRA_HOME/data/jemh/jemh-lowdiskspace.lck}}.

1. If more than 250MB is available, any historic lock file will be cleared and normal processing resumes

2. If less than 250MB is available, and no lock-file was yet created, JEMH creates the lock-file and generates an audit event, resulting in a forward notification about the problem:

   1. Open

      

   2. No mail is processed or marked read - no data loss outcome

3. The next time Jira invokes JEMH, the (current) lack of disk space and presence of the lockfile results in:

   1. No additional audit event generated, no additional forward notifications (as JEMH will be invoked possibly 000’s of times before anyone does something).

   2. No mail is processed or marked read - no data loss outcome

4. The cycle above repeats until available disk space is above the minimum, at which time, the lockfile is cleared, and normal processing resumes.

Test Cases

You can validate an instance is out of disk space by running a Test Case, it will indicate ON HOLD status, and that Mailbox Action is IGNORE (no action taken).

Open