Difference between revisions of "Tier1 Operations Report 2019-07-10"
(→) |
(→) |
||
(5 intermediate revisions by one user not shown) | |||
Line 11: | Line 11: | ||
|} | |} | ||
− | * | + | * The only issue of note to report was CMS reporting significant AM/Hammer-cloud tests and job failures. |
+ | |||
+ | After investigation it appears as result of AAA and HC tests that were in error for a number of days during the week. This was traced back to a large number of CMS production jobs (“premix” jobs ), that seemed to require significant I/O activity. (Possibly also an effect of merge jobs TBC) | ||
− | + | The strain that was put on RAL disk Storage by this load had several effects: A significant number of the Production jobs were failing (~50 or 60 %), HC tests were failing, and AAA SAM tests were also failing. | |
+ | |||
+ | For the avoidance of doubt and to aid clarity, the reason the Premix jobs were affecting the SAM tests and HC jobs so much was the following chain of events. | ||
+ | 1) Premix job would run on a worker node with heavy IO over xroot from ECHO | ||
+ | 2) Some event, possible the "slow request" interrupts the read. | ||
+ | 3) The job backs off from reading directly from ECHO and tries AAA | ||
+ | 4) AAA locates the file at RAL and the job carries on its heavy IO via the AAA proxies. | ||
+ | 5) This happens to enough jobs that it knocks the AAA proxies over | ||
+ | 6) SAM and HC errors ensue. | ||
+ | If it is possible to break the chain at link 4, this would make the system more robust. | ||
<!-- ***********End Review of Issues during last week*********** -----> | <!-- ***********End Review of Issues during last week*********** -----> | ||
Line 341: | Line 352: | ||
<!-- ************************************************************************* -----> | <!-- ************************************************************************* -----> | ||
<!-- **********************Start Hammercloud Test Report************************** -----> | <!-- **********************Start Hammercloud Test Report************************** -----> | ||
+ | |||
+ | {| width="100%" cellspacing="0" cellpadding="0" style="background-color: #ffffff; border: 1px solid silver; border-collapse: collapse; width: 100%; margin: 0 0 1em 0;" | ||
+ | |- | ||
+ | | style="background-color: #b7f1ce; border-bottom: 1px solid silver; text-align: center; font-size: 1em; font-weight: bold; margin-top: 0; margin-bottom: 0; padding-top: 0.1em; padding-bottom: 0.1em;" | Hammercloud Test Report | ||
+ | |} | ||
+ | |||
+ | {| border=1 align=center | ||
+ | | Target Availability for each site is 97.0% | ||
+ | |} | ||
+ | {| border=1 align=center | ||
+ | |- bgcolor="#7c8aaf" | ||
+ | ! Day !! Atlas HC !! CMS HC !! Comment | ||
+ | |- | ||
+ | | 2019-06-03 || 100 || n/a || | ||
+ | |- | ||
+ | | 2019-06-04 || 100 || n/a|| | ||
+ | |- | ||
+ | | 2019-06-05 || 100 || n/a || | ||
+ | |- | ||
+ | | 2019-06-06 || 100 || n/a || | ||
+ | |- | ||
+ | | 2019-06-07 || 100 || n/a || | ||
+ | |- | ||
+ | | 2019-07-09|| 100 || 45 || | ||
+ | |- | ||
+ | | 2019-07-09 || 100 || 45 || | ||
+ | |- | ||
+ | |} | ||
Latest revision as of 10:27, 10 July 2019
RAL Tier1 Operations Report for 10th July 2019
Review of Issues during the week 26th June 2019 to the 3rd July 2019. |
- The only issue of note to report was CMS reporting significant AM/Hammer-cloud tests and job failures.
After investigation it appears as result of AAA and HC tests that were in error for a number of days during the week. This was traced back to a large number of CMS production jobs (“premix” jobs ), that seemed to require significant I/O activity. (Possibly also an effect of merge jobs TBC)
The strain that was put on RAL disk Storage by this load had several effects: A significant number of the Production jobs were failing (~50 or 60 %), HC tests were failing, and AAA SAM tests were also failing.
For the avoidance of doubt and to aid clarity, the reason the Premix jobs were affecting the SAM tests and HC jobs so much was the following chain of events. 1) Premix job would run on a worker node with heavy IO over xroot from ECHO 2) Some event, possible the "slow request" interrupts the read. 3) The job backs off from reading directly from ECHO and tries AAA 4) AAA locates the file at RAL and the job carries on its heavy IO via the AAA proxies. 5) This happens to enough jobs that it knocks the AAA proxies over 6) SAM and HC errors ensue. If it is possible to break the chain at link 4, this would make the system more robust.
Current operational status and issues |
Notable Changes made since the last meeting. |
- NTR
Entries in GOC DB starting since the last report. |
Service | ID | Scheduled? | Outage/At Risk | Start | End | Duration | Reason |
---|---|---|---|---|---|---|---|
- | - | - | - | - | - | - | - |
Declared in the GOC DB |
Service | ID | Scheduled? | Outage/At Risk | Start | End | Duration | Reason |
---|---|---|---|---|---|---|---|
- | - | - | - | - | - | - | - |
- No ongoing downtime
Advanced warning for other interventions |
The following items are being discussed and are still to be formally scheduled and announced. |
Listing by category:
- DNS servers will be rolled out within the Tier1 network.
Open
GGUS Tickets (Snapshot taken during morning of the meeting). |
Ticket-ID | Type | VO | Site | Priority | Responsible Unit | Status | Last Update | Subject | Scope |
---|---|---|---|---|---|---|---|---|---|
142155 | USER | cms | RAL-LCG2 | urgent | NGI_UK | in progress | 2019-07-09 20:07:00 | Transfers are failing from UK to KIPT | WLCG |
142127 | TEAM | lhcb | RAL-LCG2 | urgent | NGI_UK | in progress | 2019-07-09 15:53:00 | 2 files cannot be staged | WLCG |
140447 | USER | dteam | RAL-LCG2 | less urgent | NGI_UK | on hold | 2019-07-09 15:03:00 | packet loss outbound from RAL-LCG2 over IPv6 | EGI |
140220 | USER | mice | RAL-LCG2 | less urgent | NGI_UK | waiting for reply | 2019-07-04 17:55:00 | mice LFC to DFC transition | EGI |
GGUS Tickets Closed Last week |
Ticket-ID | Type | VO | Site | Priority | Responsible Unit | Status | Last Update | Subject | Scope |
---|---|---|---|---|---|---|---|---|---|
141990 | USER | cms | RAL-LCG2 | urgent | NGI_UK | solved | 2019-07-09 14:11:00 | Intermittent HC failures at T1_UK_RAL | WLCG |
141968 | USER | cms | RAL-LCG2 | very urgent | NGI_UK | solved | 2019-07-04 14:20:00 | SAM (CE) and Hammer Cloud Failures at T1_UK_RAL | WLCG |
141771 | USER | cms | RAL-LCG2 | urgent | NGI_UK | closed | 2019-07-08 23:59:00 | file read error at T1_UK_RAL | WLCG |
140870 | USER | t2k.org | RAL-LCG2 | less urgent | NGI_UK | solved | 2019-07-09 14:18:00 | Files vanished from RAL tape? | EGI |
139672 | USER | other | RAL-LCG2 | urgent | NGI_UK | solved | 2019-07-09 11:57:00 | No LIGO pilots running at RAL | EGI |
Availability Report |
Day | Atlas | CMS | LHCB | Alice | Comments |
---|---|---|---|---|---|
2019-07-03 | 100 | 100 | 100 | 100 | |
2019-07-04 | 100 | 100 | 100 | 100 | |
2019-07-05 | 100 | 100 | 100 | 100 | |
2019-07-06 | 100 | 100 | 100 | 100 | |
2019-07-07 | 100 | 100 | 100 | 100 | |
2019-07-08 | 100 | 100 | 100 | 100 | |
2019-07-09 | 100 | 100 | 69 | 71 |
Hammercloud Test Report |
Target Availability for each site is 97.0% |
Day | Atlas HC | CMS HC | Comment |
---|---|---|---|
2019-06-03 | 100 | n/a | |
2019-06-04 | 100 | n/a | |
2019-06-05 | 100 | n/a | |
2019-06-06 | 100 | n/a | |
2019-06-07 | 100 | n/a | |
2019-07-09 | 100 | 45 | |
2019-07-09 | 100 | 45 |
Key: Atlas HC = Atlas HammerCloud (Queue RAL-LCG2_UCORE, Template 841); CMS HC = CMS HammerCloud
Notes from Meeting. |