Is there currently a problem with
scratch on Yggdrasil? I am trying to clone a repo (a small one, code only) and it always results in
fatal: The remote end hung up unexpectedly KiB | 5.00 KiB/s
fatal: early EOFs: 38% (1584/4167), 168.00 KiB | 7.00 KiB/s
fatal: index-pack failed
/tmp works fine.
At the same time, I am trying to pull some data (again, nothing big) from
https://kalousis.hcpdufour.unige.ch and it takes ages.
And now the problem is gone.
@support is there some ongoing work to stabilize the performance? I understand there has to be some variance depending on the current load. But currently, severs are completely unusable from time to time. Running experiments in such a randomized environment is difficult. I would even prefer having reduced top performance but with reliable access to computational resources.
One of the consequences is that one has to request 3x longer jobs to account for unpredictable performance drops when accessing/writing the data.