[00:00:49] !log admin upgrading eqiad cloudservices nodes to OpenStack Xena [00:28:34] !log tools.stewardbots SULWatcher/manage.sh restart #all bots disconnected [00:28:36] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.stewardbots/SAL [12:20:00] !log tools upgrading ingress-nginx to v1.3 [12:20:03] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [13:44:40] !log paws Upgrade ingress-nginx to 1.3.0 8e6b577fbc8a48167919dd20365967d68eb60a7d [13:44:43] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Paws/SAL [17:03:12] is it possible to add additional network interfaces to an instance? [17:03:49] define 'possible' [17:04:34] openstack may support it, but iirc we're not using extra interfaces anywhere and haven't really tested them in our setup [17:05:08] what are you trying to do? assigning multiple IPs to a single interface is something we've done before and are using in some parts of the infra, but still needs manual admin actions [17:07:29] taavi: thanks, I am testing our mail server setup in the cloud, and are existing mail servers have multiple ips, so I was hoping to replicate the setup to increase the fidelity of my testing [17:07:47] definitely not essential, but I wanted to find out if it was possible [17:09:27] pseudo-interfaces will probably serve quite well for that [17:13:03] Platonides: can I add another IP to the same interface, will our cloud route that IP? [17:19:05] hmm, that's an interesting question [17:19:20] perhaps you need a floating ip for that? [17:21:52] no, for security reasons the network will block ARP advertisements and other traffic for IPs that haven't been configured as allowed for that interface [17:22:04] we can configure extra ips manually upon request if there's a good reason [17:23:11] Platonides: it's not possible to attach a floating IP directly as an extra IP on an interface, but it is possible to map a floating IP to an extra IP with private addressing, we actually use that at least in PAWS [19:07:30] when using Cloud VPS, I am struggling with very laggy/unreliable SSH connections - am I the only one? Is there any workaround? I am following https://wikitech.wikimedia.org/wiki/Help:Accessing_Cloud_VPS_instances [20:21:21] pintoch: I would guess it has something to do with the network between you and the datacenter we call "eqiad". You could try going through the traceroute and mtr data capture steps at https://wikitech.wikimedia.org/wiki/Reporting_a_connectivity_issue to collect information for the network team folks to look into. [20:50:45] Hi 👋 [20:50:52] Nice to meet you [20:51:21] bd808: thanks! I can share a traceroute, but because I have no problems at all with Toolforge projects, I suspect the connectivity to eqiad is fine. Could it be somehow related to the "jump" setup required to access a VPS? [20:52:01] pintoch: do you access the bastion without issues? [20:52:29] and, are you doing other stuff with cloud hosts? [20:52:43] such as transferring files to/from some cloud machines? [20:53:17] I can try to do some "random stuff" (non disruptive) on bastion to compare [20:53:45] I am not doing any other stuff with those hosts at the same time. [20:54:23] comparing `ssh bastion.wmcloud.org` to `ssh ..eqiad1.wikimedia.cloud` would be a reasonable way to rule in/out the bastion as a jump host. Off the top of my head I can't think of any issues I've ever seen with jumps in general. [20:54:35] but given your replies this seems to clearly indicate that this is not a common problem, so this must be down to my VPS [20:55:38] there are people who certainly have ssh issues, but as far as I can remember they generally turn out to be general network path issues of some sort [20:56:14] ok. I will try to see if it correlates with high load on the VPS too [20:57:10] (not having a lot of CPU/RAM available is common on this instance) [20:57:30] thanks a lot for the pointers in any case! [20:58:29] If your $HOME on the target instance is on NFS that could be a source of delay too, especially if you have complex dotfiles [21:10:12] bd808: I have seen that a connection through a bastion slows down if you are transferring other kind of data (such as files) multiplexed over the same bastion connection [21:10:31] it's reasonable, though [21:10:37] sure, bandwidth is finite [22:07:10] !log admin replaced cloudservices1003 with cloudservices1005 T304888 [22:07:14] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Admin/SAL [22:07:14] T304888: Q4: (Need By: TBD) rack/setup/install 6 wmcs hosts - https://phabricator.wikimedia.org/T304888