OP Notes 6 Jul 09
From TeraGrid Wiki
(Difference between revisions)
| Revision as of 18:10, 6 July 2009 Kericson (Talk | contribs) ← Previous diff |
Current revision Jmlowe (Talk | contribs) (→Site Updates) |
||
| Line 24: | Line 24: | ||
| * NCSA: Mercury- filesystem space was getting a bit full due to normal short-term usage, normal purging cleared it. Abe/lincoln - working on some backup issues. Cobalt - filesystem being upgraded to next version today | * NCSA: Mercury- filesystem space was getting a bit full due to normal short-term usage, normal purging cleared it. Abe/lincoln - working on some backup issues. Cobalt - filesystem being upgraded to next version today | ||
| * TACC: Two outstanding tickets from Nanohub: | * TACC: Two outstanding tickets from Nanohub: | ||
| - | # https://tickets.ncsa.uiuc.edu/tickets-myproxy/cgi/ticket_display.cgi?number=173391 (Mike: This most likely means that they are nearly 7 months out of date with their TG certificates, keeping up to date with the teragrid CA certificates is a minimum requirement for interoperability.) | + | # https://tickets.ncsa.uiuc.edu/tickets-myproxy/cgi/ticket_display.cgi?number=173391 (Mike: This most likely means that they are nearly 7 months out of date with their TG certificates, keeping up to date with the teragrid CA certificates is a minimum requirement for interoperability. (Later found to not be the problem - JML)) |
| # https://tickets.ncsa.uiuc.edu/tickets-myproxy/cgi/ticket_display.cgi?number=173483 (Mike: Same errors across the board, they seem to be somehow assembling their own incorrect host DN's and fail when they don't match the real ones. This looks to be client side library abuse - see healthy looking gridftp speed page.) | # https://tickets.ncsa.uiuc.edu/tickets-myproxy/cgi/ticket_display.cgi?number=173483 (Mike: Same errors across the board, they seem to be somehow assembling their own incorrect host DN's and fail when they don't match the real ones. This looks to be client side library abuse - see healthy looking gridftp speed page.) | ||
| # Nanohub gram problem across TeraGrid (http://nanohub.org/usage/gridprobe?mode=pf&group=TERAGRID(GRAM2)&window=00086400&site=&sortKey=destination&sortDir=ascending) (Mike: Failing stage in due to gridftp client misuse/certificate maintenance accounts for nearly all of these errors.) | # Nanohub gram problem across TeraGrid (http://nanohub.org/usage/gridprobe?mode=pf&group=TERAGRID(GRAM2)&window=00086400&site=&sortKey=destination&sortDir=ascending) (Mike: Failing stage in due to gridftp client misuse/certificate maintenance accounts for nearly all of these errors.) | ||
| * PSC: Pople is down do to hardware issues. At this time we do not know when it will be back up. PSC also has a ticket open for NanoHub. Logon to PSC fail with error sent to list. | * PSC: Pople is down do to hardware issues. At this time we do not know when it will be back up. PSC also has a ticket open for NanoHub. Logon to PSC fail with error sent to list. | ||
| - | * LSU: A week ago GridFTP servers were experienced by Lisa Childers from ANL who initiated every hour about one hundred long-term gridftp (striped and nonstriped) transfers. In effect a lot of memory was allocated and other users (Inca for example) were accidentally unable to initiate their gridftp transfers. Lonestar was touched by the same problem as I checked in log files there. I suppose that more TG machines were the same problems. Results on Robbert Budden's SpeedPage would suggest it. Unfortunately there is no option to limit the number of connections per user id. | + | * LSU: A week ago GridFTP servers were experienced by Lisa Childers from ANL who initiated every hour about one hundred long-term gridftp (striped and nonstriped) transfers. In effect a lot of memory was allocated and other users (Inca for example) were accidentally unable to initiate their gridftp transfers. Lonestar was touched by the same problem as I checked in log files there. I suppose that more TG machines were the same problems. Results on Robbert Budden's SpeedPage would suggest it. Unfortunately there is no option to limit the number of connections per user id. |
| - | + | ||
| == Misc == | == Misc == | ||
Current revision
|
|
|
|
[edit]
Site Updates
- NCSA: Mercury- filesystem space was getting a bit full due to normal short-term usage, normal purging cleared it. Abe/lincoln - working on some backup issues. Cobalt - filesystem being upgraded to next version today
- TACC: Two outstanding tickets from Nanohub:
- https://tickets.ncsa.uiuc.edu/tickets-myproxy/cgi/ticket_display.cgi?number=173391 (Mike: This most likely means that they are nearly 7 months out of date with their TG certificates, keeping up to date with the teragrid CA certificates is a minimum requirement for interoperability. (Later found to not be the problem - JML))
- https://tickets.ncsa.uiuc.edu/tickets-myproxy/cgi/ticket_display.cgi?number=173483 (Mike: Same errors across the board, they seem to be somehow assembling their own incorrect host DN's and fail when they don't match the real ones. This looks to be client side library abuse - see healthy looking gridftp speed page.)
- Nanohub gram problem across TeraGrid (http://nanohub.org/usage/gridprobe?mode=pf&group=TERAGRID(GRAM2)&window=00086400&site=&sortKey=destination&sortDir=ascending) (Mike: Failing stage in due to gridftp client misuse/certificate maintenance accounts for nearly all of these errors.)
- PSC: Pople is down do to hardware issues. At this time we do not know when it will be back up. PSC also has a ticket open for NanoHub. Logon to PSC fail with error sent to list.
- LSU: A week ago GridFTP servers were experienced by Lisa Childers from ANL who initiated every hour about one hundred long-term gridftp (striped and nonstriped) transfers. In effect a lot of memory was allocated and other users (Inca for example) were accidentally unable to initiate their gridftp transfers. Lonestar was touched by the same problem as I checked in log files there. I suppose that more TG machines were the same problems. Results on Robbert Budden's SpeedPage would suggest it. Unfortunately there is no option to limit the number of connections per user id.
[edit]
Misc
- Steve Clark Nanohub cert problem. Purdue may be behind in CA certs - Von will follow up and cc Robert.
- Based on a thread a few weeks ago, Paul Brown pulled together the following KB article on decommissioning a TG resource: http://teragrid.org/cgi-bin/kb.cgi?docid=ayqc&portal=1. Please review and let Paul (cc'ed) know of any corrections.
