Good evening everyone,
After AllisonHouse installed 6.13.10 on our production servers, we noted that
on one of our machines, if you fed data from an internal server and, as a
backup, an external server…it could cause pqact processes to stop writing to
disk if the internal server's upstream LDM was restarted. Weird, huh?
Along with Steve, myself, and Daryl Herzmann from Iowa State, we found other
bugs, and Steve found several bugs on the NOAAport ingest side as well. This
bugfix also stops global warming, and prevents your smoke detector from going
off at 2 AM with a dead battery, but these latter features have not been
adequately tested.
But seriously…many of these bugs have been around for many years, and it's time
they met their demise. UNIDATA, (thank you, Steve!l Emmerson!) AllisonHouse,
College of Dupage (thank you, Mike Zuranski) and Iowa State (thank you, Daryl
Herzmann, aka Superman) hammered the beta versions, and squashed all of the
bugs we found. Not to mention, a few people found other bugs on their own, and
these got fixed as well with the new 6.13.11 release.
The download is at:
ftp://ftp.unidata.ucar.edu/pub/ldm
And a list of all of those bugs that we sent to their doom are here:
https://github.com/Unidata/LDM/blob/master/CHANGE_LOG
In the process, we found that version LDM 6.13.6, which we failed back to,
randomly crashes under Centos 7, and 6.13.10, the previous version, was stable
except for the logging bug that was fixed. However, if you have a satellite
dish with a NOAAport ingester, test immediately and upgrade ASAP. And I would
strongly recommend everyone else do the same as well, even if you are not
having any apparent issues.
Some of those bugs are serious, even if you don't see them occurring on your
servers. Those memory leaks we and Coverity Scan found are not fun…
Onward and upward!
Gilbert
