Hello all,

I am in need of isolating GET-requests referring to long-running, read-only elog functions such as search/filter/sort in our Apache proxy and redirecting them elsewhere. There does not, however, appear to be any easy way of reliably isolating these functions (with the exception of sort) by only looking at the URL.

Does anybody have any suggestions?

Regards,

Philip

I guess the underlying problem is the long time these requests take and block other users.

I have pretty high on my todo list to convert ELOG into a multi-threaded server which would fix this completely. So if you are patient enough (=months) you might get what you want.

Thanks for the quick response!

It's great to hear that multi-threading is in the works as this has been my main issue with an otherwise very nice piece of software. I do, however, feel like we should be able to get my slightly hacky approach to work to hold us over until you finish.

The idea is to run separate ELOG instances in read-only mode dedicated to these types of requests. I've managed to sync up the logbook indexation of each instance so now, unless there's some statefulness to ELOG that I'm forgetting about, I only need to make sure that requests are forwarded to the right instance.

Is there no good way of differentiating search operations from others by URL?

Look for "&subtext=" in the URL

This only applies to searches which specify that they are searching through the message text though. It would not work for things like quick filter

For any other filter you need "&<attribute>=", which of course requires the knowlede of all attributes. There is no other "standard" flag in the URL indicating a search.

Would there be a simple way to redirect all URLs BUT the ones which trigger searches?