Re: Display of bridge member interfaces cut short - bug or intention?
- In reply to: Michael Gmelin : "Re: Display of bridge member interfaces cut short - bug or intention?"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Thu, 21 Dec 2023 15:56:08 UTC
On Wed, 20 Dec 2023 21:32:45 +0100 Michael Gmelin <freebsd@grem.de> wrote: > On Wed, 20 Dec 2023 18:04:36 +0100 > "Patrick M. Hausen" <hausen@punkt.de> wrote: > > > Hi all, > > > > as some probably know we provide web hosting services and we use > > jails for that. > > > > On some particular host we have 255 vnet jails all of which are > > connected to the external interface of the host - renamed to "inet0" > > in our environment - via if_bridge(4) and all managed with iocage. > > > > root@ph003:~ # grep inet0 /iocage/jails/vpro*/config.json|wc -l > > 255 > > > > Of these 251 also have a second epair interface connected to a > > private bridge named "priv1". These are used for connections to the > > central database server which should not be exposed to the Internet. > > > > root@ph003:~ # grep priv1 /iocage/jails/vpro*/config.json | wc -l > > 251 > > > > While looking for a different problem to my great suprise I found > > today that ifconfig truncates the list of member interfaces for both > > bridge instances. And both to the same value of 102, although the > > member numbers are (albeit slightly) different: > > > > root@ph003:~ # ifconfig inet0|grep member:|wc -l > > 102 > > root@ph003:~ # ifconfig priv1 | grep member: | wc -l > > 102 > > > > All 255 jails are connected to the external network and perfectly > > reachable from the Internet. That's why I conclude that the display > > is wrong, not the bridge configuration. > > > > What's happening here? Is this intentional or shall I file a bug > > report? > > > > More importantly: either way is this only cosmetic or will we hit > > another unexpected limit of the number of interfaces that can be > > members of a bridge any time soon? > > > > Hi Patrick, > > I could reproduce the issue and created a fix: > https://reviews.freebsd.org/D43135 > > As far as I can tell the issue is cosmetic (unless, of course, you > have automation based on libifconfig or the output of the ifconfig > command). > > Best > Michael > The fix landed in main[0] and will be MFCed to releng/13.2. Cheers Michael [0]https://cgit.freebsd.org/src/commit/?id=3d36053ca6d6a17d408c8f92c504e6135dc9d8df -- Michael Gmelin