I'm back from long California trip. Here's the scoop:
Good news:
- The high amount of packet drops (I see 3 out of 10 packets drop
from South Carolina) is not due to the host site, but to network
problems at a high level ISP. (probably at MAE-East) Apparently a
lot of people, including international folks, have really good
connections. Even better, there appears to be some pressure to
solve the problem, since we aren't the only service affected.
- The host site (VAResearch) appears to be doing an excellent
job. The host machine is on a switched 10 megabit ethernet and has
some great neighbors. Best of all, it appears to be a high clue
network.
Bad news:
- Looks like two pieces of mail of the last 20,000 got misfiled
because of -generous; looks like I'll have to turn that off (good
thing it got logged) and do a manual cleanup.
- I still haven't reached my goal of having nothing show up in the
trouble log for a week. I think the record is around 48 hours.
Changes:
* "rebuild" (a manual tool for cleaning up messes) was made slightly
more user friendly. Nobody will ever see the effects.
* The search engine indexing was enhanced (again) to be a bit more
incremental. Indexing time dropped by 30%, and a pass just now
took 2 hours, 23 minutes. Consequently, I switched to nightly
indexing runs instead of semi-weekly ones. The search for a list
engine should be a bit more up to date, plus we can back off if
need arises.
* I added a few minor sanity checks on the names of lists we
archive; things like making sure there are no quotes in a
listname, etc.
* Ultra minor cosmetic code cleanup.
That's a wrap, tune in next time for "what's going on in the world
of mail-archive.com". I suspect we are getting ever so closer to
a critical mass judging by recent growth.
Jeff
