Pile of Dross

I’ve finally decided to merge the two blogs I’ve been keeping for a while, the personal one on LiveJournal and the work one on my own server. So as to not throw the few people who actual read my LJ one off, and not fill the LJ namespace with a million and one RSS feeds, I’m using Live-press to synchronise the two. On the upside, there’ll be more entries, as I try and update the work one everyday. On the downside, the work ones tend to be a brain dump written 10 minutes before I leave the office and are normally fairly techy. Currently I’m not intending to merge older entries

More fiddling with selinux, syslog-ng can now read from /proc/kmsg but now can’t open up files on /pool, and neither could syslog when I tried that.

Edited /etc/selinux/targeted/src/policy/filecontexts/types.fc to add /pool as a var_log_t type but that apparently made no difference.
Eventually discovered using ls -dZ that /pool had no security context defined, so set it using

setfattr -n "security.selinux" -v "system_u:object_r:var_log_t\000" /pool

then rebooted the system to check it survived a reboot, which it did. Then tried to start syslog-ng, failing with similar error, checked file it was talking about- it was one I’d touched earilet and so it had the default security context, rm’d it and syslog-ng recreated it with the correct context so its all working.

Now need to check that logrotate does the right thing.

• Took delivery of :
  • A 6-channel A/V switcher hub to replace the push button one I bought when I got fed up of swapping scart leads about between Graeme’s DVD player and my PS2.
  • A USB Joy convertor(according to the invoice anyway), you see I’ve been unable to use my joy for sometime and I hoped this would let me, but sadly the joy converter only works with analogue joy not my digital joy. Bah.
  • A Logitech MX518 mouse to replace my Intellimouse Explorer
  • A Razer exactMat, though I’m not using the wrist rest as it isn’t particualrly comfortable in certain positions.
  • A Creative Audigy 4
  • A Harmony Remote 688 as my laziness is reaching new heights.
  • A Sony DVPNS30 DVD Player
• Watched the Firefly DVD boxset (twice)
• Started playing Hitman:Contracts
• Started playing Thief 3
• Completed Hitman:Contracts (with a worringly large number of levels completed with a rating of Mass Murderer)
• Gained 4 levels in World of Warcraft
• Got my Wireless AP to finally forward on DHCP requests - apparently it wouldn’t unless it had received its address via DHCP *rolls eyes*

Cleaned up remianing files from SC3 tape test - deleted path entries so files could nove to tape
Saw that atlas were trying to transfer files into tape side, checked that files were moving to tape okay, they weren’t so asked JW to look at configuration, which he fixed.
Reading up on SELinux, to understand the changes I need to make to csflnx266 to get syslog-ng working
Answering questions on GridPP-Storage list

Holiday 8th August - 12th August

Examining logfiles to try to see why David Colling, CMS @ Imperial is seeing failed transfers to our SRM, so for the examples he’s provided have not had any entry in the dcache2 database on pg350, which suggests that they aren’t reaching the SRM. I’m wondering if this is a problem with different bdii’s being at different version number and so are unable to talk to each other. I’ve mentioned it to Steve to look at while I’m on holiday

Done some more work on csflnx266 - becoming critical old syslogger makin weird noises.

Worked with Phillippa to understand why Tickets from our helpdesk weren’t getting into Footprints, tunred out replies were going out with a different e-mail address to the address footprints had sent to, changed the outgoing e-mail on the Tier1a-lcg queue.
Attended SC conference call
Mailed Andrew and Jeremy SC3 problem report
Added Chris Kruk to sg CVS
Running more transfer tests in loopback - it appears that we have problems with 32 transfers in and out, whether this 32 transfers of one type or 64 in total is unknown so far.
Ran pnfs-poolcheck on CMS area after nfs42 issure- 11 files not in pool,but still need to find and report to CMS

Continued to monitor tests, with 32 files 1 stream, get the same rate (possibly slower) for 16 files but with double the load, alos staring to see some failures, some appear to happen arounf the time I made fts changes, others don’t

Had a look at the source for lcg-infosites to see if I can work out why things are going wrong, I suspect its to do with some duplicated entries in the bdii - teras.saranl for example.

Setup up a test channel for Lancaster to Lancaster in FTS.

(Trying to do this more frequently/regularly)

-Running some test using fts and our SRM in “loopback”, with 32 files and “0″ streams seeing 1Gb/s consistent rates, with peaks up to 1.4Gb/s
-

lcg-infosites --vo X se | grep gridpp

reports weird results, investigating
-Restart myproxy service as it was playing up.
-Checked bdii with script mentioned on LCG-ROLLOUT looking for bad info system entries.