2025-03-13 Meeting Notes
Date
Mar 13, 2025
Participants
@Thomas, Jyothish (STFC,RAL,SC)
@Alexander Rogovskiy
@James Walder
@Katy Ellis
@Thomas Byrne
Emmanuel
Lancs: Gerard, Matt, Steven
Glasgow:
@Ian Johnson
Apologies:
CC:
Goals
List of Epics
New tickets
Consider new functionality / items
Detailed discussion of important topics
Site report activity
Discussion topics
Current status of Echo Gateways / WNs testing
Recent sandbox’s for review / deployments:
Item | Presenter | Notes |
|
---|---|---|---|
Operational Issues |
| Worker Node writable XCache fixed and deployed in lhcb nodes Ceph sn upgrade finished, gws next week - Quincy LSST voms authdb changes and DUNE token auth implemented. svc20 got in an 'interesting state': |
|
Mini-DC Observations, Issues, outcomes |
| Final Observations |
|
Compilation and rollout status of RAL XRootD versions | @Thomas, Jyothish (STFC,RAL,SC) | 5.7.3 released (awaiting other changes to gateways) streamed checksums to be deployed tomorrow. |
|
XRootD collaboration Meeting |
|
|
|
cms-aaa naming convention | @Thomas, Jyothish (STFC,RAL,SC) | cms-aaa is the only remaining personality to use proxy/ceph as the xrootd service names Separate naming convention would be more appropriate, to have main/supporting (not so urgent). CC created, and sandbox is prepared and has been tested on a test host |
|
cms-aaa jemalloc use | @Thomas, Jyothish (STFC,RAL,SC) | testing on svc20, some memory leak still present. added memory limits, currently under observation |
|
Shoveler | @Katy Ellis |
|
|
On the fly Checksums | @Ian Johnson
| Mis-matched checksums during mini-DC: many of these due to wrongly recording a uninitialised streamed checksum value (looking like “0*[24]*0*”) when a client requested reading a file. Code wasn’t ensuring checksum calculation only when writing data. Hence, many streamed values for a pathname (recorded in plaintext) did not match the checksum stored in the extended attribute.
|
|
Deletions | To check deletion timing split between client/cluster response under DC saturation
|
| |
XRootD Writable Workernode Gateway Hackaton
| @Thomas, Jyothish (STFC,RAL,SC)
| Being deployed this week. LHCb jobs from 2018-2021 gens are already writing their output data to ECHO via root (i.e. local gateways). Looks good so far. |
|
Plan: file query system to summarize XRootD Logs |
| Plan to create a system to store info from across all gateways to search a filename and get creation time, last write time, last successful stat and deletion time in case of ‘lost’ files. Possible graduate sideproject. Ian plans to extend the database schema from the deletion tests (capturing file write completions and deletions) into a more general event schema. |
|
100 GbE Gateway testing: | @James Walder @Thomas, Jyothish (STFC,RAL,SC) | UKSRC - Acting as source for SRCNet verification tests; not being stressed so far … Tier-1 .
|
|
UKSRC Storage Architecture |
| Tom B. Working on CephAdm setup for the cluster. JW attempting to reinstall the hosts. |
|
Tokens Status |
|
|
|
test stress testing framework |
| Script ready, first test to be arranged |
|
on GGUS:
Site reports
Lancaster:
Our Ceph seems to have recovered after our big outage. Shhh, don’t do anything to spook it.
(discussed earlier in the meeting) Whilst doing some changes in our authdb for LSST it seems to be like “lrw” permissions don’t allow a user to write, but setting permission to “a” (all) for a path does. Waters are muddied by lsst trying to be posh with groups and roles (and not allowing me to get a certificate without a group attached).
Any known issues with this? It’s been a long time since I’ve looked at this, but I swear it used to work.
Glasgow -
Action items