Am 15.07.2015 um 13:43 schrieb João Jerónimo:
On 15-07-2015 11:21, Reindl Harald wrote:However, the main question is: why the hell did the openvas installation get bricked just because one of the processes was killed, and how can I recover from this without reinstalling the whole guest OS? Isn't it the job of openvas-check-setup to find this kind of problems?no it's not the job of "openvas-check-setup" to fix bricked installations with a completly unknown state after random processes got killed by the kernel in the middle of operations - no software can fix thatMy understanding is that this "completely unknown state" is stored in de filesystem in the form of files, so it isn't "completely random" after all
it is - only god knows which processes where killed at what point in time and what half written data was there in that exactly moment
If openvas stops working and even after a reboot the problem persists then this must mean that some kind of corruption happened to openvas-related files as a result of the crash... And isn't it the job of openvas-check-setup to check openvas-related files, after all?
not in any case
I understand that a simple script can't check and fix everything (some details must be hidden deep inside thousands of binary files), but in that case don't tell me that the purpose of openvas-check-setup if not to find this kind of problems.
it's job is to find *missing configuration* but not random corruption
Normally software can tollerate crashes and don't need a reformat to fix problems. That's the general rule
try a database server with a full disk...BTW: you are the only one starts talking about re-format and reinstall the complete OS, everybody else would purge /var/lib/appname, backup it before and restore as much as possible data from that backup like the certs and so on
there is *no general rule* what happens when mutiple processes may write related and connected data to different files and they get *randomly* and *repeatet* killed by the kernel and the state you face is the result of multiple faults on your side:
* complete wrong ressource assignment * fire up a murderjob without expierience and look at the ressource usage of a smaller stop when begin to work with new SW * don't make a backup before * in case of a VM don't take a snapshot before
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Openvas-discuss mailing list [email protected] https://lists.wald.intevation.org/cgi-bin/mailman/listinfo/openvas-discuss
