STATUS updates
Some links of interest:
A lot of time has passed since my last update… not that it hasn't been without activity. The LAIR moving, the LAIR getting set back up… merely just a lapse in my documenting habits.
I am currently on an automation kick, to try and reduce the amount of information that gets pushed into my sphere of attention. Specifically non-actionable information (like normally running systems with adequate amounts of free diskspace). I only want to be notified if a normally running system is no longer normally running, and in this specific context, if its free diskspace reaches at least 80% utilization.
Aside from sokraits/halfadder, which still need a new pruning script, only lab46 and log were reaching saturation. log especially so.
lab46 was simply a matter of clearing out /root of unnecessary files (.cabal, an apparent haskell module repository) and purging a no-longer-used kernel.
log I ended up rolling some rather large and long-running log files. Compressing them (-9) really did the trick.
Seems cron wasn't entirely happy with my new scripts, and wasn't overly happy in general (was having trouble contacting the ldap server). I removed the auth3 entry and restarted it, and fixed various path issues in the troubled scripts… things seem to be okay now.
My various class metrics scripts, data, and graphs got some extra love… I now have late submissions reporting, along with class-centric metrics pages allowing for an overall trends graph being (manually) generated.
I'll have to automate that at some point…
Perhaps not quite worthy of 'ng', but I did take some time to polish and comment idler, and attempted a first draft at bringing a (now that I think about it, incomplete) weekly idle time reporting. But at least it has been given some attention, and the logic re-contemplated (even if not immediately changed).
I actually should pull some data from the 'gn gather' records being studiously kept… or do more of a week-by-week offset sort of thing.
While investigating a (once again) break week offset bug, I discovered the source of the problem… break week offsets were only being recognized if the span of days was exactly 5. My break weeks were entered including weekend dates, so they were always larger than 5. A '-ge 5' seemed to do the trick nicely.
I've been wanting to further optimize and automate some things in cron, chief among them process_attendance, opusjournal, and the various graphmetrics scripts being fired off by my neat chain of at jobs.
projtasks simplifies the whole at job thing, eliminating them and being run once per day. It is aware of not only what projects a class has, but what projects are currently active and in need of seeing metrics page refreshes.
After various tweaks, I seem to have gotten a number of my other scripts cooperating with cron as well.
While I was digging in other stuff, I rediscovered a set of holiday themed motd's I had deployed a number of years ago (say, 2011)… a few at jobs and it is now in progress.
I had some more time to continue tweaking the new podmaster image… packages installed, and the start of configuration.
Seems that login_ldap is causing the boot time hangs on daemon start. I may have to explore rearranging them, as things were notably faster before it was brought online.
OpenBSD 5.8 came out! I grabbed the ISO and set it up for netbook via the syslinux 'memdisk' module… just like booting from CD, only over the network.
Saves some time from having to manually get the right files in place, although base sets detection seems a little off.