There is a pending feed update. The scan can not be started

Hi There,

In gvm-11; I notice a scan won’t run if a feed update is pending. Problem, in case of a scheduled scan the task will be set in error and won’t resume unless manually restarted after the feed update is done.

Is it possible to have scheduled tasks wait for the feeds update to finish then automatically (re)start ?

Typical scenario is in case you have Task 1 running, with an alert condition set to start Task 2 when Task 1 si finished. If in the middle a feed update is run, Task 2 won’t start and will be set in error.

Thanks !

Currently this isn’t possible.

You MUST NOT start a scan while a feed update is running and you MUST NOT update the feed while a scan is running. Both can result in unwanted and unspecified behavior. With the next release (current master branches) we will even forbid both situations.

1 Like

Thanks bricks, but then how do you sequence this properly ? If you have scheduled scans you don’t knows when they are suppose to finish, and since you still need to have your feeds manually updated (thus, at fixed time)… eg, the more tasks you have scheduled, the higher risk you take to have this situation happens.

Unless there is a way for gvm to update the feeds at idle time or something like this ?

I noticed the feed updates happen late in the morning at irregular times.

There is a file “plugin_feed_info.inc” in the plugins directory which tells you the timestamp of the new feed after you downloaded it.

So once you figure out the time rage (observing or even logging that info for some days), you could schedule your scans after the “latest” feed update. Which of course keeps being an estimate as long as you don’t benefit of any guaranteed publication times and downloading rates. (Don’t know if such a thing exists.)

Well that’s wobbly. Don’t know what the plan is for the future, but this should definitely be managed by gvmd automatically, with all tasks put on hold until this is finished. I don’t see how you can make this work smoothly when you have hundreds of different targets, multiple users having access to the plateform, and scheduled scans.

Out of curiosity, was this also the case with gvm-9/OpenVAS ? I don’t recall having this problem before.

Thanks