Public bug reported: We've got a cluster of diskless compute nodes that we use for highly CPU-intensive tasks. To keep them minimal, they're headless and diskless, and they PXE-boot and run NFS-root from a central server. We recently upgraded the PXE-boot kernel for these nodes to an Ubuntu 16.04 series kernel. Specifically, uname -a on the nodes returns:
Linux fluorine 4.4.0-83-generic #106-Ubuntu SMP Mon Jun 26 17:54:43 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux After the upgrade, we started noticing that anytime the NFS link got heavily loaded, and the NFS server started getting bogged down trying to keep up with the requests, the diskless clients would spontaneously unmount their /proc filesystem. It sounds an awful lot like this report: https://forums.gentoo.org/viewtopic-t-1010040-start-0.html ...the fix for which was supposedly integrated into the mainline kernel as of 3.19.2. But we never saw this issue until just recently, right after upgrading to 4.4.0. Did the patch that fixed this in 3.19 get somehow reverted in 4.4? Or was a new trigger for this bug somehow introduced? ProblemType: Bug DistroRelease: Ubuntu 16.04 Package: linux-image-4.4.0-83-generic (not installed) ProcVersionSignature: Ubuntu 4.4.0-97.120-generic 4.4.87 Uname: Linux 4.4.0-97-generic x86_64 ApportVersion: 2.20.1-0ubuntu2.10 Architecture: amd64 AudioDevicesInUse: USER PID ACCESS COMMAND /dev/snd/controlC1: chaz 2928 F.... pulseaudio /dev/snd/controlC0: chaz 2928 F.... pulseaudio CurrentDesktop: XFCE Date: Fri Nov 3 14:26:29 2017 InstallationDate: Installed on 2014-07-22 (1200 days ago) InstallationMedia: Xubuntu 14.04 LTS "Trusty Tahr" - Release amd64 (20140416.2) MachineType: Micro-Star International Co., Ltd. GT70 2PC ProcFB: 0 inteldrmfb 1 nouveaufb ProcKernelCmdLine: BOOT_IMAGE=/vmlinuz-4.4.0-97-generic root=UUID=9dda5f94-8c6c-45a4-8a4c-38b3f69a8b15 ro rootfstype=btrfs rootflags=discard,ssd,compress=no,subvol=@ nosplash RelatedPackageVersions: linux-restricted-modules-4.4.0-97-generic N/A linux-backports-modules-4.4.0-97-generic N/A linux-firmware 1.157.12 SourcePackage: linux UpgradeStatus: Upgraded to xenial on 2016-08-08 (452 days ago) dmi.bios.date: 04/17/2015 dmi.bios.vendor: American Megatrends Inc. dmi.bios.version: E1763IMS.71D dmi.board.asset.tag: To be filled by O.E.M. dmi.board.name: MS-1763 dmi.board.vendor: Micro-Star International Co., Ltd. dmi.board.version: REV:0.C dmi.chassis.asset.tag: To Be Filled By O.E.M. dmi.chassis.type: 10 dmi.chassis.vendor: Micro-Star International Co., Ltd. dmi.chassis.version: To Be Filled By O.E.M. dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvrE1763IMS.71D:bd04/17/2015:svnMicro-StarInternationalCo.,Ltd.:pnGT702PC:pvrREV0.C:rvnMicro-StarInternationalCo.,Ltd.:rnMS-1763:rvrREV0.C:cvnMicro-StarInternationalCo.,Ltd.:ct10:cvrToBeFilledByO.E.M.: dmi.product.name: GT70 2PC dmi.product.version: REV:0.C dmi.sys.vendor: Micro-Star International Co., Ltd. ** Affects: linux (Ubuntu) Importance: Undecided Status: New ** Tags: amd64 apport-bug xenial ** Attachment added: "proc_version and lspci from affected diskless client" https://bugs.launchpad.net/bugs/1729993/+attachment/5003214/+files/debug_traces_from_fluorine.txt -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1729993 Title: proc umounts on nfsroot client when nfs server is heavily loaded To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1729993/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
