Hello,
we have a SRX3400 cluster configured with "set system archival configuration transfer-on-commit". Now I realized that the secondary node was not able to transfer its config file to the configured host for some time.
We now have a lot of config files in /var/transfer/config/ which could not be transferred. If I repair the configuration, all these file would be transferred, which is not necessary. So I would like to delete a big part of all those files.
I am now wondering how this archival job is working and if it is safe to delete files manually from that directory. I haven't found information on this. I currently can think of two possibilities:
1. The archival job checks from time to time if there are files in that directory. If yes, they are transferred in an order from oldest to newest. Once a file is transferred, the job deletes the file. I think, in this case it would be safe to delete the files manually from the directory.
2. Upon commit a new config file is created for transfer and a list is updated with the new file name. The archival job parses this list and transfers the listed files. Once the file is transferred, its name is removed from the list and the file is removed as well.
If I delete files manually, I would not update this list. Maybe it makes a difference, if I delete those files from shell (rm -f /var/transfer/config/$filename_pattern) or from CLI (file delete /var/transfer/config/$filename_pattern)?
Does someone have a hint on this?
Thanks in advance,
Hermod