Closed bernd closed 1 year ago
More detailed steps to reproduce in a lab:
Prereqs: mmdb.zip (link redacted)
/etc/graylog/server/
mv
or cp
commands
/etc/graylog/server/
Workaround:
Deleting the file (via rm
) and then doing a cp/mv appears to prevent this from happening.
@Graylog2/secdev Assuming that SecDev could take this, could we get a high priority for this one?
We had recent issues in Cloud related to geo ip, and I would like to prevent more frustration if possible.
When this has been solved and we have a timeline for when this will be patched in Cloud, can someone please PM me and also put a note confirming such on HS-961472138.
I request the same update for HS-1004828525
Hi, we also have another customer experiencing the same issue. Please see HS-999920232
Thanks a lot for working on this! We would like to roll this out to Cloud as quickly as possible.
Can we get a backport to 4.3 please?
/cc @craig-thomas
@boosty @craig-thomas working on it now, we should have a backport PR in by tomorrow!
@kingzacko1 Thanks a lot 🙌
Expected Behavior
The GeoIP message processor (GeoIpResolverEngine class) should reload the database files when they change.
Current Behavior
The
GeoIpResolverEngine
loads the database files only once at startup and doesn't re-open the database handles when a database file gets updated.Possible Solution
Introduce a reload mechanism similar to how MaxMindDataAdapter#refresh works.
The code should also handle in-place database file updates correctly and gracefully handle any errors. (see Context below)
Steps to Reproduce (for bugs)
Context
@drewmiranda-gl noticed the missing reload behavior while updating GeoIP databases. The code raised the following exception when a database file got replaced with the
cp
command. We should handle such cases and reload the database handle.Your Environment