[04:02] <MACscr> any ideas what im doing wrong here in regards to dnsmasq and resolv? it shows the correct ip's when i check the dnsmasq status, but then says no servers found in /var/run/dnsmasq/resolv.conf. Is that required if its getting them fine through my network interface file? https://paste.debian.net/hidden/1b71f30c/
[04:02] <MACscr> this is a xenial lts server
[05:14] <MACscr> weird. rebooted system to apply kernel updates and everything is good again
[05:40] <Ouyes> guys, is there any way that I can limit the buffer/cache used on ubuntu server?  My server is using too much memory for cache/buffer, the application is running lack of memory, sometimes it failed to allocate memory.
[08:22] <lordievader> Good morning
[08:24] <lordievader> Ouyes: Buffers/cache should be reclaimable memory by an application. You can drop them, but they are there for your benefit.
[08:24] <lordievader> Ouyes: https://www.tecmint.com/clear-ram-memory-cache-buffer-and-swap-space-on-linux/
[08:31] <Ouyes> lordievader, well  let me tell you something, the buffer/cache is used and controlled by the kernel, when free memory is low, the application will report "can't not allocate memory".
[08:33] <lordievader> "Disk cache can always be given back to applications immediately! You are not low on ram!" from https://www.linuxatemyram.com/
[08:33] <lordievader> So, buffers/cache should not be a problem. But like I said, you can drop them.
[08:35] <Ouyes> lordievader, drop them ? how?  the application should got the priority to get memory, and the kernel should maintain how much should be put into cache, so in your opinion, every time when an application wants to get some memory, it should free some cache?
[08:37] <lordievader> The application does get the priority. The kernel only uses memory for cache/buffers that is unused. The kernel only needs to free some cache if there is no 'free' memory available.
[08:39] <Ouyes> lordievader, this is exactly the problem, the kernel did not free cache when application needs it.
[08:39] <Ouyes> lordievader, never mind, I increased my swap memory.
[08:41] <lordievader> Ouyes: Do you have `sysstat` installed? What is the output of `sudo sar -r 1 1`. The memory might simply be higly overcommitted.
[08:43] <Ouyes> lordievader,  overcommitted?
[08:44] <lordievader> Yes, there is a difference between how much memory an application request v.s. how much it actually uses.
[08:44] <lordievader> The kernel, by default, allows overcommitting till 150% if I recall correctly.
[08:45] <Ouyes> lordievader,  I am not quite following you, what are u trying to indicate? my application has a memory leakage?
[08:46] <Ouyes> lordievader, it is actually a ubuntu server
[08:47] <lordievader> Ouyes: No, that is not what I'm trying to say. Could you run the command I gave you?
[10:45] <ahasenack> good morning
[13:35] <samba35> if i want usb drivers to be loaded as a module ,not kernel componet then how do i remove usb drivers from kernel and add to as a module
[13:37] <rbasak> kstenerud: php7.3 should be availab.e in git-ubuntu now.
[13:50] <kstenerud> great Thanks!
[15:18] <muhaha> any idea how to set iptables permanently ? (18.10)
[15:18] <sdeziel> muhaha: check the netfilter-persistent package
[15:22] <Greyztar> With rsync,if sync job is rather large and to be most efficient would it be ok to do for the initial sync with --partial(-p),then if it gets scuffed by disconnect,one would use the --append-verify,then for further syncing use the --checksum option?
[15:33] <sdeziel> Greyztar: you could always run it with --partial, no? --checksum is rather expensive for large files since you need to read them all on both sides
[15:35] <Greyztar> sdeziel: yes that checksum hmm,but wouldnt partial start a whole new transfer again if i just redo that command after a disconnect or so?
[15:35] <sdeziel> Greyztar: no, --partial is specifically made to keep the partial file around in case of disconnect
[15:37] <sdeziel> Greyztar: what I don't like with --append-verify is the implied --inplace and the side effects it has
[15:37] <sdeziel> Greyztar: but maybe those side effects are OK for your use case, I don't know :)
[15:37] <Greyztar> sdeziel: thanks for the heads up i think i need to look further into this,ive only used abit gluster and ceph which for me atleast was set and forget hehe
[15:38] <sdeziel> Greyztar: for large file transfer, I really love zfs send/receive feature
[15:39] <Greyztar> sdeziel: zfs ive yet to handle,my dream is to have my storage all on zfs,but for homelab i find it not so flexible with raidz when cant add remove was it device or vdev ?
[15:42] <Greyztar> sdeziel: though they say they will implement it,when they do ill migrate,think can actually remove device now though,but add not so much
[15:42] <sdeziel> Greyztar: I've only used mirrors, sorry
[15:43] <Greyztar> sdeziel: ohh ok then it doesnt matter,but i fully agree zfs is the way to go thumbs up
[16:51] <Greyztar> sdeziel: did some reading though,would it be that omit append with --inplace as to if source file could get corrupted so have like 3 versions to compare with?
[16:55] <smoser> rbasak: katamo and Odd_Bloke (and a customer) were confused as to why bug 1802354 was not fixed in bionic.
[16:56] <smoser> i think the source of that was (partially) the message in comment 17
[16:56] <smoser> where would i request a fix to tooling that posted that message to mention "bionic-updates"
[16:56] <rbasak> smoser: see pending-sru
[16:56] <smoser> rather than just "-updates"
[16:56] <rbasak> There's a resoruce-agents armhf dep8 failure
[16:56] <rbasak> smoser: ubuntu-archive-tools
[16:56] <rbasak> smoser: sru-review script
[16:57] <rbasak> (also possibly sru-accept script)
[16:57] <smoser> ok.
[16:58] <smoser> katamo, Odd_Bloke i will follow up on trying to improve that message
[16:58] <katamo> smoser I'll try to read more carefully too ;) thanks for that follow through though!
[16:58] <smoser> but one of you should follow up in #ubuntu-release on requesting that failure to be ignored
[16:59] <smoser> http://autopkgtest.ubuntu.com/packages/r/resource-agents/bionic/armhf
[16:59] <smoser> katamo and for reference, "pending sru" that rbasak mentioned was
[16:59] <smoser>  https://people.canonical.com/~ubuntu-archive/pending-sru.html
[18:36] <v0lksman> I have an l2tp connection I'm trying to establish.  I believe I've set everything correctly however when I connect I don't see a new interface or IP bind to any of the existing interfaces. However in syslog I do see keep alives being sent and a few other entries regarding the tunnel
[18:36] <v0lksman> wondering if the remote (the "server") should be serving me an IP at this point or is this something I need to set strongswan or xl2tpd to receive?
[18:38] <v0lksman> I'm asking in server because the answers I've gotten to date are to use a GUI tool that was created, but I can't as it's headless
[18:49] <v0lksman> all good...bad config
[18:57] <leftyfb> Anyone know the proper replacement for Wants=network-online.target? Since this doesn't actually work.
[18:57] <leftyfb> In a systemd unit that is
[18:58] <sdeziel> leftyfb: have you also set After=network-online.target?
[18:58] <leftyfb> I've tried that as well
[18:58] <leftyfb> and Requires
[18:59] <leftyfb> and every permutation of those
[18:59] <leftyfb> I've seen lots of posts online of network-online not actually working the way it's supposed to
[19:00] <sdeziel> systemctl is-enabled NetworkManager-wait-online.service systemd-networkd-wait-online.service => do you have any of those "enabled"
[19:00] <leftyfb> No
[19:01] <sdeziel> or maybe you are using ifupdown?
[19:01] <leftyfb> it's a server so NetworkManager isn't installed
[19:01] <leftyfb> it's 16.04, so yes, ifupdown
[19:01] <rbasak> leftyfb: do you know about https://www.freedesktop.org/wiki/Software/systemd/NetworkTarget/ /
[19:01] <rbasak> ?
[19:01] <sdeziel> I think you could depend on ifup@$IFACE or something
[19:02] <leftyfb> rbasak: yep. Read the entire page several times. It's stating to do what we already had configured originally.
[19:03] <sdeziel> I'd try Wants= and After=ifup@$IFACE.service
[19:03] <rbasak> I'm not sure how well ifupdown is integrated with network-online.target.
[19:04] <leftyfb> sdeziel: we have a gross workaround using "until", but I was looking for a cleaner solution
[19:04] <rbasak> The page explains why network-online.target is ill-defined
[19:04] <leftyfb> rbasak: right, so is there a better alternative?
[19:06] <rbasak> leftyfb: for what? http://xyproblem.info/ etc
[19:06] <leftyfb> alternative to network-online.target
[19:07] <leftyfb> to determine if we're online .
[19:09] <rbasak> Define "online"!
[19:09] <rbasak> See the upstream page :)
[19:13] <leftyfb> ping able to contact a host over the internet ... in this case, a REDIS server
[19:13] <teward> i'm going to go out on a limb and say that "online" is a determination of if the network interface is up
[19:14] <teward> not necessarily a connection to "other machines"
[19:14] <teward> if you are looking for an "actually able to reach things" state I'm not sure there's a SystemD target that fits that requirement
[19:15] <leftyfb> network-online is supposed to mean interface is up, has an ip and can contact a DNS server. <~~~ this would be adequate, but it doesn't work
[19:16] <teward> ehhh... "network-online.target is a target that actively waits until the nework is "up", where the definition of "up" is defined by the network management software. "  <-- this is the actual definition
[19:16] <teward> according to rbasak's link
[19:16] <teward> but that's not really a 'clear' definition
[19:16] <teward> Or more specifically: "This will ensure that all configured network devices are up and have an IP address assigned before the service is started."  <-- this
[19:17] <teward> this doesn't mean that it'll verify connectivity exists if I'm reading this right
[19:17] <teward> just that it's got an IP assigned
[19:17] <leftyfb> that would be good if that were the case
[19:17] <rbasak> Unfortunately whatever definition you pick it will not work for some reasonably substational set of server users.
[19:17] <teward> ^ this
[19:17] <teward> which is why it's not really in 'mainstream' use :p
[19:17] <rbasak> That's why the best answer is as suggested in that page - fix the services.
[19:17] <teward> ^ this
[19:17] <leftyfb> ok, so there's no systemd solution
[19:18] <teward> it'd be better for your app/service to have in-built ability to test its own connectivity and 'not start' in those cases where it can't reach the other servers.
[19:18] <sdeziel> I don't even know if this target has any real meaning when using ifupdown
[19:18] <lordcirth> Yeah, the app (or a launch wrapper) should check
[19:18] <lordcirth> You could use ExecStartPre= for that, perhaps
[19:20] <leftyfb> lordcirth: Again, this is what we're already doing as a workaround. I was just looking for a cleaner way. Looks like there isn't one.
[19:20] <sdeziel> leftyfb: have you tried the Wants/After=ifup@$IFACE.service?
[19:20] <sdeziel> this worked for me in the past
[19:22] <leftyfb> ExecStartPre=/bin/bash -c 'until host remotehost.com; do sleep 1; done'
[19:22] <leftyfb> This is the workaround we're using. It works
[19:23] <rbasak> An "up" ifupdown stanza would be better than that.
[19:23] <lordcirth> Oh ok, sorry.  Didn't read the whole scrollback
[19:23] <rbasak> Or sdeziel's suggestion sounds even better
[19:28] <leftyfb> sdeziel: That's cleaner. I like it. Thank you
[19:29] <sdeziel> leftyfb: you are welcome
[21:03] <DammitJim> is there such a thing as openjdk 10?
[21:04] <nacc> DammitJim: generally? Yes, but not in any current release, I don't think
[21:05] <DammitJim> whatever happened to it? got superceded by 11?
[21:05] <nacc> https://launchpad.net/ubuntu/+source/openjdk-lts/+publishinghistory
[21:05] <nacc> gives you a rough idea, i suppose
[21:06] <nacc> DammitJim: so that package builds openjdk 11, which confusingly for 18.04 users is still at 10.x .. It will update at some point in 18.04 (I believe there is an open bug for this you canfind easily)
[21:07] <nacc> DammitJim: LP: #1796027
[21:07] <DammitJim> OMG
[21:07] <DammitJim> thanks for clarifying