v3.210601

use cmsweb-preprod, not cmwseb-test2 to test parasitic upload to S3
since test clusters are firewalled and only accessible inside CERN

v3.210505

add parasitic uploading of tarballs to S3 CRABCacheDev

v3.210422

• Make crab uploadlog work with s3 (fix #4971 and fix #4993)

v3.210420

make also submit --dryrun work with S3 cache

• -submit --dryrun can use S3. Fix #4991 (#4992)

v3.210416

small but critical fix for S3:

• tarballname has to end with .tar.gz Fix #4988 (#4989)

v3.210413

forgot t5 import one fix in yesterday's tag.

• use latest S3 utils functions from ServerUtilities v3.210410

v3.210412

BUG FIXING for S3 cache

• fix computation of hashkey (tarballname) for sandbox when using S3 #4985

without this, all sandboxes where having same name :-(

v3.210409

this first step toward a Client compatible with using S3 as crabcache.
Needs CRABServer v3.210409 and WMCore 1.3.6.crab6

crab submit works with S3 but TaskWorker is not S3-capable yet.
log upload is not there yet.

Hopefully this still works on current prod/testbed, but extensive testing is needed.

v3.210323

minor fixes for a better life, especially the one about not publishing multiple datasets

• refuse to submit if asked to publish multiple datasets. fix #4980 (#4982)
• improve messages when task bootstraps. fix #4977 (#4981)
• remove annoying message from crab status about dashboard

v3.210127

some small fixes and a large change to instantiate CRABServer REST object in a single place so to add proper user-agent header CRABClient/versionnumber and send all communications with CRABServer on cmsweb* via port 8443

NOTE: needs to be built with CRABServer v3.210127 or later

See #4969 (comment) for an initial list of things to test

List of changes:
-adapt all commands to use a common RESTServer instance (#4969)
-explicitely set userAgent to CRABClient (#4969)
-fix #4967 (#4968)
-try rucio.lfns2pfns with write first, than fallback to read (#4964)
-Create tmp crab3checkwrite file only if lfn2pfn was successful (fix #4961) (#4962)