August 2012 - Core-students - mailman3.cels.anl.gov

When bones go snap.. Status update
by Gene Rackow 03 Aug '12

03 Aug '12

So for those that haven't heard, here's the details... On Weds Jul 25, I was turning from Lemont Road into the parking lot for Hollywood Blvd theator. The lite had been yellow a bit and it really looked like the oncoming car was slowing down to stop at the light. Looking to the lot as I then made the turn, the next thing I hear is squealling tires and I'm laying on what would be the sidewalk of the entranceway. For some reason my right leg was taking a strange direction between the knee and ankle. Shoulder hurt as well. My guess is that the oncoming car didn't see me and the motorcycle so tried to speed up to make the light as it turned red. He then hit the brakes about the same time as he hit me. Just speculation. A few minutes later the ambulance arrived and took me off to Good Sam. A few hours of surgery later I now have a shiny new titanium rod and several screws inthe right leg. The right clavical/shoulder is broken, but not in a way that needs surgery. No weight on either for some unknown amount of time. I've been told the padding in the back side of the helmut is fairly well compacted too. Without that, I probably wouldn't be capable of writting this. On Tues they moved me to SilverCross hospital rehab unit. Delays in scheduling, not health caused the delay. The initial claim was about 2 weeks there to figure out how to stand, move, etc with only 1 side working. It's not as easy as you might think. Esp when it's the left side that's still working. Things are going well so they cut that down to 1 week. I should be getting out of here early next week. That will be release to home, not back to work. No weight on right side and lots and lots of rest needed. Wheelchair, no cane or walker for some amount of time. I'm doing well, but bored to tears with olympics and bad tv. Reading tends to put me to sleep, always has. That sleep doesn't really help though. Oh well.. --Gene

2 1

Where're John[BV]?
by John Valdes 03 Aug '12

03 Aug '12

[taking the liberty of speaking for JohnB as well :) ] Will be attending the Infiniband User Day(s) in the TCS Conference Center this Mon and Tue, Aug 6 and 7. Will be onsite and likely checking on things, depending on how interesting/relevant the talks are... John

1 0

Where's Craig 8/7
by Craig Stacey 03 Aug '12

03 Aug '12

I forgot to mention I'll be at Searle on Tuesday (since I'll be missing my Monday out there). -- Craig

1 0

Where's Craig 8/6
by Craig Stacey 03 Aug '12

03 Aug '12

I have an all-day Doctor's appointment on Monday. Please put me down for one sick day. Thank you. -- Craig

1 0

Today's user termination list
by stace＠mcs.anl.gov 03 Aug '12

03 Aug '12

Full list of people who no longer appear on the active employee list =========================================================================== 51615 PSC Quock, Deborah E.R. 53071 PHY Jisonna, Louis J. 56211 PHY Mohr, William H. 58736 PHY O'Shaughnessy, Christopher 64772 PSC Tanner, John J. 64821 PSC Guzman Lopez, Javier 66474 PSC Lee, Jonathan R. I. 67401 PSC Lepesheva, Galina I. 84504 PSC Vaidya, Anand T. 84815 PSC Wang, Jiyang 88187 PSC Macfie, Sheila M. 203900 PSC McCahill, Jenny S.J. 205218 PSC Mohr, Juergen Andreas 216343 PSC Dolan, Daniel H. 217557 PSC Dhatwalia, Richa 218250 PHY Moerland, Daniel S. 219489 PSC Bandla, Sudheer 219840 PSC Kim, Jin Seok 219844 PSC Kim, Ji Ae 220174 PSC Bartolome Usieto, Jose Fernando 220184 PSC Johannes, Ludger 221035 PSC Megharaj, Prabhu 221040 PSC Carvalho, Tobi 221074 PSC Aoyagi, Kenji 221102 PSC Sun, Qingxiang 221126 PSC Jennings, Eleanor S. 221150 CEP Purdie, Michael 221151 PSC Fleming, Samantha E. 221152 PSC Dobbins, Margaret M. 221155 PSC Bane, Lukas B. 221158 PSC Baur, Robin M. 222721 PHY Watson, Kristin A. 224827 ES Huynh, David 225450 NE Pichler, Benjamin K. 225686 DIS Ruan, Joseph Y.

1 0

Transferring SEED filesystems from one solaris system to another [LONG]
by Hunter Matthews 02 Aug '12

02 Aug '12

00. We had issues with root access to both the affected seed file servers. I've nuked the issue from orbit by adding warehouse13's root key to them and tested it -both work. a. ssh to namby or pamby - warehouse13 only allows logins from these hosts or the console b. login to warehouse13.mcs.anl.gov as root. The password is in an obvious location in craigs office. It should be the same one as the one in the safe as well. c. You are root in warehouse13 - ssh sto02.mcs.anl.gov ssh sto09.mcs.anl.gov should now make you root. 0. Facts needed - which filesystem by zpool name is moving and from which server and to which server. Recommendations: use 2 windows, one with the source and target logged in as root. 1. Make sure the add on utility called "mbuffer" is installed in the PATH on _both the source and target machines. (The current situation is with the SEED servers sto02 and sto09 and I have already installed it there. /usr/bin/mbuffer and its tested and working). 2. Create an initial SNAPSHOT on the source side. For the data-servers filesystem on sto02 this becomes: zfs snapshot dpool/data-servers@aug2 Note: I (hunter) typically use the partial date of snap creation for these transfer snapshots. Many filesystems have autogenerated snaps and I never use these. 3. Create an mbuffer instance to receive this filesystem on the target. For sto09 as the target this becomes: mbuffer -s 128k -m 10G -I 9092 | zfs receive -v dpool/data-servers Note 1: 9092 is the port number and they simply have to match and not be used Note 2: -s and -m are the buffering options These are overkill, but work well. Note 3: we don't list the snapshot name on the TARGET, IE leave off the "@aug2" part Note 4: this command will block, waiting for the input to start. 4. Using zfs send, send it through another mbuffer instance. To send the snapshot created above to our waiting target, this is: zfs send dpool/data-servers@aug2 | mbuffer -s 128k -m 10G -O sto09.mcs.anl.gov:9092 Note 1: mbuffer on both sides will start showing traffic immediately basically. Note 2: zfs send/recieve will always transfer more data than the listed size of the filesystem. 10-15% is typical. Don't panic. 5. Once thats finished, mark the source filesystem as read only. (prevents changes via NFS while we do the 2nd phase of the copy) zfs set readonly=yes dpool/data-servers 6. Do a new final snapshot of the source zfs snapshot dpool/data-servers@final 7. Again, create an mbuffer instance to receive on the target: This is EXACTLY the same command as last time: mbuffer -s 128k -m 10G -I 9092 | zfs receive -v dpool/data-servers 8. Do an incremental snapshot send. This only sends the diffs from the first snapshot (step 2 and step 6 above). zfs send -i dpool/data-server@aug2 dpool/data-servers@final | mbuffer -s 128k -m 10G -O sto09.mcs.anl.gov:9092 Note 1: should be considerably faster than last one. 9. Start removing the source filesystem - on SEED servers this can take a LONG time. zfs destroy -r dpool/data-servers 10. Change NIS (they still have some systems using NIS, I think) a. login to bio-admin-1 and become root b. edit the /etc/auto.vol map - look for ALL the places that reference data-servers I found two. Change the server to sto09.mcs.anl.gov - everything else stays the same. c. cd to /etc/mcs and run ./UPDATE d. say a small prayer it works this time. 11. Remembering the filesystems found in NIS, make similar changes to LDAP. a. For testing and NFSv4 reasons, SEED has 4 sets of maps in LDAP under ou's starting with 'seed' (seed, seedtest, seedbackup, seednfsv4) b. ldapsearch -x -b dc=mcs,dc=anl,dc=gov -LLL cn=data-servers >> change_me ldapsearch -x -b dc=mcs,dc=anl,dc=gov -LLL cn=kmer-server-prod >> change_me c. Edit that file - I made all the entries look like the following GENERALIZED example dn: cn=data-servers,ou=auto.vol,ou=mounts,ou=seedbackup,dc=mcs,dc=anl,dc=gov changetype: modify replace: automountInformation automountInformation: -fstype=nfs,rw,nosuid,nodev,vers=3,rsize=1048576,wsize=1048576,hard,intr,noacl,proto=tcp,timeo=600,retrans=2 sto09.mcs.anl.gov:/dpool/& 12. Apply the 8 changes (2 names across 4 ou's = 8 changes) with ldapmodify a. kinit SOME_ADMIN b. ldapmodify -f change_me 13. That _should_ be it. Some SEED systems delight in hanging onto old NFS mounts, so you'll have to beat on those or just reboot them. 14. In addition, I have NO IDEA how to make nfsv4 go - so if magellen doesn't magically work against the new server, I don't know how to fix it. -- Hunter Matthews Unix Administrator Office: Bldg 240 Room 2139 Argonne National Labs, MCS Key: F0F88438 / FFB5 34C0 B350 99A4 BB02 9779 A5DB 8B09 F0F8 8438 Never take candy from strangers. Especially on the internet.

1 1

Transferring ifs filesystems from one solaris system to another
by Hunter Matthews 02 Aug '12

02 Aug '12

0. Facts needed - which filesystem by zpool name is moving and from which server and to which server. Recommendations: use 2 windows, one with the source and target logged in as root. 1. Make sure the add on utility called "mbuffer" is installed in the PATH on _both the source and target machines. (The current situation is with the SEED servers sto02 and sto09 and I have already installed it there. /usr/bin/mbuffer and its tested and working). 2. Create an initial SNAPSHOT on the source side. For the data-servers filesystem on sto02 this becomes: zfs snapshot dpool/data-servers@aug2 Note: I (hunter) typically use the partial date of snap creation for these transfer snapshots. Many filesystems have autogenerated snaps and I never use these. 3. Create an mbuffer instance to receive this filesystem on the target. For sto09 as the target this becomes: mbuffer -s 128k -m 10G -I 9092 | zfs receive -v dpool/data-servers Note 1: 9092 is the port number and they simply have to match and not be used Note 2: -s and -m are the buffering options These are overkill, but work well. Note 3: we don't list the snapshot name on the TARGET, IE leave off the "@aug2" part Note 4: this command will block, waiting for the input to start. 4. Using zfs send, send it through another mbuffer instance. To send the snapshot created above to our waiting target, this is: zfs send dpool/data-servers@aug2 | mbuffer -s 128k -m 10G -O sto09.mcs.anl.gov:9092 Note 1: mbuffer on both sides will start showing traffic immediately basically. Note 2: zfs send/recieve will always transfer more data than the listed size of the filesystem. 10-15% is typical. Don't panic. More notes coming. Is this readable/comprehensive enough? -- Hunter Matthews Unix Administrator Office: Bldg 240 Room 2139 Argonne National Labs, MCS Key: F0F88438 / FFB5 34C0 B350 99A4 BB02 9779 A5DB 8B09 F0F8 8438 Never take candy from strangers. Especially on the internet.

2 1

where's Linda
by Linda Winkler 02 Aug '12

02 Aug '12

8/2 Lab until 4pm 8/3 VACATION 8/6 Lab 8/7-8 travel/Denver SC12 Exec Comm 8/9-12 PTO in Denver/SLC 8/13-14 travel/SLC SC12 SCinet meeting 8/15 travel day/SLC-ORD 8/16 Urbana for MREN Exec Comm meeting 8/17 Lab 8/20-23 UIC for Ciena 6500 training lw

1 0

Where's JohnR?
by John Roberts 02 Aug '12

02 Aug '12

Congestion is making it difficult to even see and function today. Since tomorrow will be a lighter day on systems I'm going to take today as a sick day and make it back tomorrow. Sorry for the late notice, was hoping I could ignore it. Sent from my iPhone

1 0

where's Linda 8/3
by Linda Winkler 02 Aug '12

02 Aug '12

It turns out that my half-day meeting on Tues was a full day. I will be recouping the vacation time and taking Fri as a full day of vacation. lw

1 0