General information
Middleware
UMD 4.8.5
- "DPM-dedicated" release to support migration to 1.13.0 (see dedicated section)
- release candidate ready, integration tests in progress, TBR very soon
Preview repository
- released on 2019-08-16
- Preview 1.24.0 AppDB info (sl6): APEL Client/Server 1.8.1, APEL-SSM 2.4.0, ARC 6.1.0, davix 0.7.4, dCache 4.2.40, DMLite/DPM 1.13.1, Dynafed 1.5.0, frontier-squid 4.8.1, gfal2 2.16.3
- Preview 2.24.0 AppDB info (CentOS 7): APEL Client/Server 1.8.1, APEL-SSM 2.4.0, ARC 6.1.0, davix 0.7.4, dCache 4.2.40, DMLite/DPM 1.13.1, Dynafed 1.5.0, frontier-squid 4.8.1, gfal2 2.16.3
Operations
ARGO/SAM
FedCloud
Feedback from DMSU
Known Error Database: https://wiki.egi.eu/wiki/KEDB
Tickets handled by DMSU:
- SRM argo/nagios probes GGUS 142022: Decision on preferred storage access protocols, required monitoring probes (assigned to Operations, on hold)
- SRM probes need to be updated since they are still using the old lcg-utils which have been replaced by GFAL2
- considering the SRM end of support for DPM, the reference protocol will be HTTP/webdav
- CREAM cannot initialize connection to BLAH BLParserClient GGUS #142621:
- CREAM/BLAH supports only single version of HTCondor (8.6.3) currently, does not work with 8.8.4. No fixes for HTCondor 8.6 line are expected anymore since 8.8 is more than six months old already. CREAM team may be able to respond in case of actual emergency (required changes could be very small) but has not commented yet.
- BDII udpate & base64 encoded ldif entries GGUS #142928:
- bdii-update does not handle base64-encoded values in LDIF correctly, non-ASCCII characters or passwords invoking mandatory base64 use are not present in BDII information normally, but bdii-update should handle it anyway (developers have not commented yet).
- CREAM-CE at CentOS 7: could not create connection to database server GGUS #142425 (SOLVED):
- Puppet recipe for CREAM-CE requests presence of DB server, but does not handle its configuration fully (even the parameters set by Puppet on the CREAM side), manual checking and adjustment may be necessary
ERROR org.glite.ce.commonj.db.DatasourceManager - Cannot create PoolableConnectionFactory Could not create connection to database server. Attempted reconnect 3 times. Giving up.
- solved by setting in /etc/glite-ce-cream/cream-config.xml:
url="jdbc:mysql://localhost:3306/..." (instead of url="jdbc:mysql://lgdce01.jinr.ru:3306/...")
- changes in VOMS server certificate GGUS #142627, GGUS #142820:
- whene there is a change of DN of VOMS server's certificate, all the clients need to update their configuration
- OpenStack-VM-ops probe failed since automatic image deletion GGUS #142905:
- AppDB entry of VM image used by monitoring probe has expired and manual refresh was necessary, the probe does not check (and issue warning in advance) for this itself (regular maintaineer of the image was on vacation), likely repeat yearly
Notifications from ARGO about the nagios probes failures
In the process of implementing the notification system in ARGO, it was introduced the following changes in GOC-DB:
- Notifications flag at the site level
- Notifications flag at the service endpoint level
In this way ARGO will retrieve from GOC-DB the information about the sites and services whom sending the email notification and the related recipients.
The logic of the notifications is the following:
Site | Service | Notify? |
---|---|---|
Y | Y | Y |
Y | N | N |
N | Y | N |
N | N | N |
If there isn't any email contact defined at the service endpoint level, it will be used the site contact. Please review your contacts.
You can enable the notifications.
NOTE: It is not mandatory for the sites
Monthly Availability/Reliability
- Under-performed sites in the past A/R reports with issues not yet fixed:
- AsiaPacific: https://ggus.eu/index.php?mode=ticket_info&ticket_id=142591
- INDIACMS-TIFR: SRM service not published in the BDII. No feedback yet
- TW-NTU-HEP: SRM certificate test failures; performance recovered
- NGI_AEGIS: https://ggus.eu/index.php?mode=ticket_info&ticket_id=140555
- AEGIS01-IPB-SCL: CE host certificate renewed; SRM not yet published in the BDII;
- NGI_GRNET: https://ggus.eu/index.php?mode=ticket_info&ticket_id=140556
- GR-12-TEIKAV: SRM instable; some investigations are on-going
- NGI_IT: https://ggus.eu/index.php?mode=ticket_info&ticket_id=142592
- INFN-LECCE (recovered)
- INFN-MILANO-ATLASC
- NGI_PL: https://ggus.eu/index.php?mode=ticket_info&ticket_id=140557
- TASK: still problems: SRM failures and QCG bug;
- NGI_RO: https://ggus.eu/index.php?mode=ticket_info&ticket_id=142160
- RO-02-NIPNE: Cooling system problem, migration to CentOS7
- NGI_UA: https://ggus.eu/index.php?mode=ticket_info&ticket_id=142161
- UA-IFBG: site reinstalled after a downtime caused by power interruption in the underlying data center. In August the performance were perfect, since September problems with ARC-CE
- NGI_UK:
- 100IT https://ggus.eu/index.php?mode=ticket_info&ticket_id=142163 eu.egi.cloud.OCCI-VM test returns unknown status. Waiting for completing the switch to openID connect.
- AsiaPacific: https://ggus.eu/index.php?mode=ticket_info&ticket_id=142591
- Under-performed sites after 3 consecutive months, under-performed NGIs, QoS violations: (Aug 2019):
suspended sites: AEGIS11-MISANU (NGI_AEGIS), IMBG (NGI_UA),
IPv6 readiness plans
- INFORMATION ARE 6 MONTHS OLD, please provide updates to the IPv6 assessment (ongoing) https://wiki.egi.eu/w/index.php?title=IPV6_Assessment
- if any relevant, information will be summarised at OMB
LCGDM end of support and migration to / enabling of DOME
- The DPM team has agreed to extend support for security updates for the DPM legacy functionality until 30 September 2019. However, affected service providers should still plan to disable legacy mode well before this date.
- more details: https://wiki.egi.eu/wiki/DPM_End_of_legacy-mode_support
- since DPM 1.10.3 release, it is possible enabling the non-legacy mode DOME (Disk operations Management Engine, see documentation)
- latest DPM release 1.13.1 (to be included in UMD)
- EGI Operations sent a broadcast to site-admins(first bunch and second bunch) and VO managers with a survey to fill in:
- site-admins: to know the upgrade plans: https://www.surveymonkey.com/r/2PBZSNB
- VO Managers: to know any barriers in moving away from using SRM protocol https://www.surveymonkey.com/r/2ZMZJVG
- Deployment statistics: 85 sites (100 servers)
$ ldapsearch -x -LLL -H ldap://bdii.marie.hellasgrid.gr:2170 -b "Mds-Vo-Name=local,o=grid" '(&(objectClass=GlueSE)(GlueSEImplementationName=DPM))' GlueSEImplementationVersion | grep -i ^glue | sort | uniq -c 14 GlueSEImplementationVersion: 1.10.0 21 GlueSEImplementationVersion: 1.12.0 2 GlueSEImplementationVersion: 1.12.1 22 GlueSEImplementationVersion: 1.13.0 17 GlueSEImplementationVersion: 1.8.10 2 GlueSEImplementationVersion: 1.8.11 2 GlueSEImplementationVersion: 1.8.7 4 GlueSEImplementationVersion: 1.8.8 3 GlueSEImplementationVersion: 1.8.9 14 GlueSEImplementationVersion: 1.9.0
- all the sites with older DPM versions than 1.12 are suggested to upgrade to the latest DPM version , following the guide DPM upgrade (chapter 1 Upgrade to DPM 1.10.0 "Legacy Flavour" and chapter 2 Upgrade to DPM 1.10.0 "Dome Flavour")
- DOME and the old LCGDM (srm protocol) will coexist
- the upgrade will be followed-up mainly by WLCG
- EGI Operations will liaise with the non-WLCG sites by opening GGUS tickets after DPM 1.13 is reeleased into UMD
- Monitoring: sites should enable the monitoring of the HTTP/WebDav endpoint
- register the storage service endpoint as WebDav service type, with production flag disabled, providing the URL field (see the HOWTO21
- check if the tests are ok
- switch the production flag to "yes"
HTCondorCE integration
Link to procedure: https://wiki.egi.eu/wiki/PROC19
GGUS ticket: https://ggus.eu/index.php?mode=ticket_info&ticket_id=139377
Steps status:
- (Completed) Underpinning Agreement:
- (Completed) Configuration management:
- Service type already present: org.opensciencegrid.htcondorce
- (To do) OPS Dashboard:
- To do when Conf Mgmt and Monitoring are completed.
- (in progress) Information System: the info-provider is already available: https://github.com/opensciencegrid/htcondor-ce/tree/master/contrib/bdii
- to test during the staged rollout
- (in progress) Monitoring: some probes are already available that submit jobs to CondorCe, CREAM and ARC-CE:
- https://gitlab.cern.ch/etf/jess
- ticket to ARGO to deploy the probes on the test instance: https://ggus.eu/index.php?mode=ticket_info&ticket_id=141177
- (Completed) Support: the support unit is https://wiki.egi.eu/wiki/GGUS:HTCondor-CE_FAQ
- (in progress) Accounting:
- Some work has been done: https://twiki.cern.ch/twiki/bin/view/LCG/HtCondorCeAccounting
- ticket https://ggus.eu/index.php?mode=ticket_info&ticket_id=139403 to track the progress on this topic
- APEL 1.8.0-1 adds support for a cputmult scaling factor.
- Liverpool site managed to publish accounting records
- To involve more (UK) sites in the testing
- INFN-T1 joined the tests
- problems should have been fixed with APEL 1.8.1
- (Completed) Documentation
- (Completed) Security
- Condor team filled in the questionnaire and sent it to Linda
- (in progress) UMD
- suggested to act in advance because the functional tests could take a lot of time
- Ticket for the inclusion: https://ggus.eu/index.php?mode=ticket_info&ticket_id=139764
- HTCondor-CE is released with HTCondor in a version without the dependencies on other OSG software. Moreover, HTCondor-CE depends on HTCondor (though not for use as a batch scheduler), so both packages will need to be distributed.
- Asked for any automated deployment (ansible, puppets or other) configuration files supported by HTcondor team
- It is available a copy of the ansible files that OSG uses to configure HTCondor-CE on their internal testbed machines. They can provide help for a non-OSG version of the files
- Provided information for the UMD card: https://wiki.egi.eu/wiki/UMD_products_ID_cards
AOB
Next meeting
October https://indico.egi.eu/indico/event/4782/