I’m trying to use PUMA2 to run UKCM2, suite id u-ea310. Although the PUMA2 setup seems to be running ok, it gets submit-failed for all tasks that it tries to send to ARCHER2 (see screenshot).
From the error message in the /home/n02/n02/radiam24/cylc-run/u-ea310/run1/log/scheduler/log file it looks like you’ve got a wonky cylc environment which is part cylc8 part cylc7.
This error message is the command help from Cylc7!
2026-07-07T15:16:51Z ERROR - platform: archer2 - initialisation did not complete
COMMAND:
ssh -oBatchMode=yes -oConnectTimeout=8 \\
-oStrictHostKeyChecking=no ln02 env CYLC_VERSION=8.6.3 \\
CYLC_ENV_NAME=cylc-8.6.3-2 bash --login -c \\
'exec "$0" "$@"' cylc remote-init archer2 \\
$HOME/cylc-run/u-ea310/run1 \\
run='$DATADIR/cylc-run/u-ea310/run1'
RETURN CODE:
2
STDERR:
Usage: cylc \[task\] remote-init \[--indirect-comm=ssh\] UUID RUND
(This command is for internal use.)
Install suite service files on a task remote (i.e. a \[owner@\]host):
.service/contact: All task -> suite communication methods.
.service/passphrase: Direct task -> suite HTTP(S) communication only.
.service/ssl.cert: Direct task -> suite HTTPS communication only.
Content of items to install from a tar file read from STDIN
Return:
0:
On success or if initialisation not required:
- Print SuiteSrvFilesManager.REMOTE_INIT_NOT_REQUIRED if initialisation
not required (e.g. remote has shared file system with suite host).
- Print SuiteSrvFilesManager.REMOTE_INIT_DONE on success.
1:
On failure.
Arguments:
UUID UUID of current suite server process
RUND The run directory of the suite
Actually, think I might see the problem. On ARCHER2 your .bash_profile is loading a really old version of our um environment and thus picking up cylc 7 instead.
On ARCHER2 in ~/.bash_profile
Change . /work/y07/shared/umshared/bin/rose-um-env
to . /work/y07/shared/umshared/bin/rose-um-env-puma2
Thanks, I just made this change in ~/.bash_profile. However, when I log out of ARCHER2 and puma2 and log back in again, even though ~/.bash_profile is changed as you said, on puma2 when I run cylc --version I still get 7.8.12.
Is there something else I need to do to make sure I’m loading CYLC 8? Do I need to delete the old .bash_profile~ and .profile~ which are the only things I can find that still point to . /work/y07/shared/umshared/bin/rose-um-env ?
You need to export CYLC_VERSION=8 on PUMA2 either on the command line or in your .profile/.bash_profile. Sorry, I assumed you’d done this to have submitted the suite in the first place as it started to run with cylc8.
Unfortunately various tasks are failing to submit again – even though I do still have ROSE_VERSION=2 and CYLC_VERSION=8. Might the issue now be with ARCHER2? I’m attaching example error logs. Any ideas?
Hopefully you subsequently, got the following alert from ARCHER2:
Dear ARCHER2 Users,
User access to ARCHER2 has temporarily been removed. This means that you will not be able to connect to the login nodes or access data. Running and queued jobs will continue to run. We will provide an update by 5pm tomorrow, Friday 10th July.
We apologise for the inconvenience caused.
