henningw left a comment (kamailio/kamailio#4787)
> > I do not see the real benefits (maybe not yet), at least for address table
> > that I use to worth such a large change that is not easier to review and
> > test for stability. Also, for subnet matching, there is also match the
> > longest one, which means going over all records, so hash table vs list
> > makes no difference.
>
> @miconda regarding the address table, it was requested by your right here:
> [#4539
> (comment)](https://github.com/kamailio/kamailio/pull/4539#issuecomment-3786639567)
>
> > Anyhow, I do not think the matching speed was an issue so far, and
> > regarding reloads, from the design of the PR I think that lookups can keep
> > using an old table while another process (e.g., rpc) frees it.
>
> This is not true, multiple reloads at a time create a race, and this was
> explained well in details in the older version of PR here: [#4539
> (comment)](https://github.com/kamailio/kamailio/pull/4539#issuecomment-3785133654)
>
> > if table-lock is acquired by each process using the table, then the
> > bucket-lock should not be necessary, but then only one process can use the
> > table at that moment, practically matching cannot be done in parallel
> > anymore
>
> This is not a true and I described this well in the commit:
> [4ebb47d](https://github.com/kamailio/kamailio/commit/4ebb47d1adc10b5eb2a6a2dfc95249e6cfc51365)
>
> > if table-lock is NOT acquired by each process using the table, then
> > reloading in parallel can result in races, with a process still accessing
> > an old table that is freed by another one
>
> The table level lock (the upper lock) is acquired always by each process who
> has to do either with reload or hr timer, and I wrote about it as well in my
> commits bodies..
>
> I see that most of what I wrote in commits bodies wasn't ever read. I abandon
> the PR.
>
> BR, Donat
Thank you @zenichev for taking the time and creating two versions of the PR.
Maybe somebody else like to pick it up and continue working on it.
--
Reply to this email directly or view it on GitHub:
https://github.com/kamailio/kamailio/pull/4787#issuecomment-4857936454
You are receiving this because you are subscribed to this thread.
Message ID: <kamailio/kamailio/pull/4787/[email protected]>
_______________________________________________
Kamailio - Development Mailing List -- [email protected]
To unsubscribe send an email to [email protected]
Important: keep the mailing list in the recipients, do not reply only to the
sender!
