The previous tests were executed in a lxc VM, but using the ubuntu-25.04-beta-live-server-amd64.iso as a starting point, instead of the lxd provided one.
The main difference between the two images is the swap memory. The server image gets swap memory by default which allows the kernel to swap out unused memory pages. The lxd image does not have swap memory so the kernel can not swap out pages when the requested virtual memory is more than memory.max. This difference can bee seen in the message buffer of the kernel when the OOM killer kills the process: # Server iso [ 397.347689] leak-memory invoked oom-killer: gfp_mask=0xcc0(GFP_KERNEL), order=0, oom_score_adj=0 [ 397.347708] CPU: 1 UID: 1000 PID: 1470 Comm: leak-memory Kdump: loaded Not tainted 6.14.0-13-generic #13-Ubuntu [ 397.347717] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009)/LXD, BIOS unknown 2/2/2022 [ 397.347727] Call Trace: [ 397.347736] <TASK> [ 397.347742] show_stack+0x49/0x60 [ 397.347770] dump_stack_lvl+0x5f/0x90 [ 397.347784] dump_stack+0x10/0x18 [ 397.347788] dump_header+0x48/0x1be [ 397.347803] oom_kill_process.cold+0xb/0xaf [ 397.347809] out_of_memory+0x100/0x2b0 [ 397.347820] mem_cgroup_out_of_memory+0x13b/0x170 [ 397.347834] try_charge_memcg+0x40f/0x5c0 [ 397.347846] charge_memcg+0x34/0x80 [ 397.347851] __mem_cgroup_charge+0x2d/0xa0 [ 397.347860] alloc_anon_folio+0x219/0x480 [ 397.347875] do_anonymous_page+0x148/0x480 [ 397.347882] handle_pte_fault+0x1db/0x200 [ 397.347889] __handle_mm_fault+0x3d2/0x7a0 [ 397.347902] handle_mm_fault+0xef/0x2d0 [ 397.347909] do_user_addr_fault+0x2ff/0x7e0 [ 397.347926] exc_page_fault+0x85/0x1e0 [ 397.347952] asm_exc_page_fault+0x27/0x30 [ 397.347960] RIP: 0033:0x5eb47b03eaf4 [ 397.347965] Code: e8 71 f7 ff ff 48 89 45 d0 48 8b 55 d8 48 8b 45 d0 48 89 10 48 8b 45 d0 48 83 c0 08 48 89 45 e0 eb 10 48 8b 45 e0 48 8b 55 c0 <48> 89 10 48 83 45 e0 08 48 8b 45 b8 48 83 e0 f8 48 89 c2 48 8b 45 [ 397.347969] RSP: 002b:00007ffc8ea0d730 EFLAGS: 00010287 [ 397.347980] RAX: 0000722fb0c0b000 RBX: 00007ffc8ea0d8f8 RCX: 0000722fb0be6010 [ 397.347986] RDX: 0000000101a00000 RSI: 0000000000100000 RDI: 0000722fb0be6010 [ 397.347988] RBP: 00007ffc8ea0d790 R08: 00000000ffffffff R09: 0000000000000000 [ 397.347990] R10: 0000000000000022 R11: 0000000000000246 R12: 0000000000000001 [ 397.347993] R13: 0000000000000000 R14: 00005eb47b040d30 R15: 00007230b3875000 [ 397.348003] </TASK> [ 397.348008] memory: usage 51200kB, limit 51200kB, failcnt 904335 [ 397.348029] swap: usage 4194052kB, limit 9007199254740988kB, failcnt 0 [ 397.348033] Memory cgroup stats for /user.slice/user-1000.slice/[email protected]/app.slice/run-p1470-i1471.scope: ... # LXD image [ 434.148492] leak-memory invoked oom-killer: gfp_mask=0xcc0(GFP_KERNEL), order=0, oom_score_adj=0 [ 434.148509] CPU: 1 UID: 1001 PID: 1240 Comm: leak-memory Not tainted 6.14.0-13-generic #13-Ubuntu [ 434.148515] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009)/LXD, BIOS unknown 2/2/2022 [ 434.148519] Call Trace: [ 434.148522] <TASK> [ 434.148527] show_stack+0x49/0x60 [ 434.148537] dump_stack_lvl+0x5f/0x90 [ 434.148544] dump_stack+0x10/0x18 [ 434.148564] dump_header+0x48/0x1be [ 434.148569] oom_kill_process.cold+0xb/0xaf [ 434.148576] out_of_memory+0x100/0x2b0 [ 434.148582] mem_cgroup_out_of_memory+0x13b/0x170 [ 434.148592] try_charge_memcg+0x40f/0x5c0 [ 434.148601] charge_memcg+0x34/0x80 [ 434.148605] __mem_cgroup_charge+0x2d/0xa0 [ 434.148611] alloc_anon_folio+0x219/0x480 [ 434.148619] do_anonymous_page+0x148/0x480 [ 434.148626] handle_pte_fault+0x1db/0x200 [ 434.148633] __handle_mm_fault+0x3d2/0x7a0 [ 434.148646] handle_mm_fault+0xef/0x2d0 [ 434.148653] do_user_addr_fault+0x2ff/0x7e0 [ 434.148663] exc_page_fault+0x85/0x1e0 [ 434.148670] asm_exc_page_fault+0x27/0x30 [ 434.148675] RIP: 0033:0x5f93dfd6daf4 [ 434.148680] Code: e8 71 f7 ff ff 48 89 45 d0 48 8b 55 d8 48 8b 45 d0 48 89 10 48 8b 45 d0 48 83 c0 08 48 89 45 e0 eb 10 48 8b 45 e0 48 8b 55 c0 <48> 89 10 48 83 45 e0 08 48 8b 45 b8 48 83 e0 f8 48 89 c2 48 8b 45 [ 434.148684] RSP: 002b:00007ffd89fb0f70 EFLAGS: 00010287 [ 434.148688] RAX: 00007ed2f6d5b000 RBX: 00007ffd89fb1138 RCX: 00007ed2f6ccf010 [ 434.148691] RDX: 0000000003200000 RSI: 0000000000100000 RDI: 00007ed2f6ccf010 [ 434.148693] RBP: 00007ffd89fb0fd0 R08: 00000000ffffffff R09: 0000000000000000 [ 434.148695] R10: 0000000000000022 R11: 0000000000000246 R12: 0000000000000001 [ 434.148698] R13: 0000000000000000 R14: 00005f93dfd6fd30 R15: 00007ed2fa1b2000 [ 434.148708] </TASK> [ 434.148710] memory: usage 51200kB, limit 51200kB, failcnt 40 [ 434.148736] swap: usage 0kB, limit 9007199254740988kB, failcnt 0 [ 434.148741] Memory cgroup stats for /user.slice/user-1001.slice/[email protected]/app.slice/run-p1240-i1241.scope: ... -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2105866 Title: cgroup: weird calculation for memory limit To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2105866/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
