On 6/24/20, Greg 'groggy' Lehey <grog at freebsd.org> wrote:> On Wednesday, 24 June 2020 at 9:36:23 -0700, Donald Wilde wrote: >> All, >> >> I recently upgraded my 12-STABLE system to the latest, and now my >> swap subsystems aren't working. I deliberately set up a 40GB >> partition for swap, and when I do 'top -t' I am only seeing 7906M >> total. > > That looks suspiciously like the difference from 32 GB. Could it be > numeric overflow? And if so, where? What does pstat -s say? >Well, hi Greg! LTNT2! pstat -shm: /dev/ada0s1b 65536 (1M blocks), Used: 1.5G, Avail: 63G, Capacity: 2% This is while synth is still running on openblas and openjdk8.>> When using synth, it shows swap usage of < 2%, but then it gives the >> 'out of swap' complaint and tosses all my users out into neverland. > > Unless this also causes kernel messages to that effect, this, too, > could be an issue with synth.I believe the messages about 'out of swap' were from the kernel as it died, not synth. As I say, it killed all my active users, both root and non-root. Why does _top_ show the 7906M number? I suspect that it's deeper than just synth. What else can I share to help diagnose this?> > Greg > -- > Sent from my desktop computer.[snip] -- Don Wilde **************************************************** * What is the Internet of Things but a system * * of systems including humans? * ****************************************************
On 6/24/20, Donald Wilde <dwilde1 at gmail.com> wrote:> On 6/24/20, Greg 'groggy' Lehey <grog at freebsd.org> wrote: >> On Wednesday, 24 June 2020 at 9:36:23 -0700, Donald Wilde wrote: >>> All, >>> >>> I recently upgraded my 12-STABLE system to the latest, and now my >>> swap subsystems aren't working. I deliberately set up a 40GB >>> partition for swap, and when I do 'top -t' I am only seeing 7906M >>> total. >> >> That looks suspiciously like the difference from 32 GB. Could it be >> numeric overflow? And if so, where? What does pstat -s say? >> > > Well, hi Greg! LTNT2! > > pstat -shm: > > /dev/ada0s1b 65536 (1M blocks), Used: 1.5G, Avail: 63G, Capacity: 2% > > This is while synth is still running on openblas and openjdk8. > >>> When using synth, it shows swap usage of < 2%, but then it gives the >>> 'out of swap' complaint and tosses all my users out into neverland. >> >> Unless this also causes kernel messages to that effect, this, too, >> could be an issue with synth. > > I believe the messages about 'out of swap' were from the kernel as it > died, not synth. As I say, it killed all my active users, both root > and non-root. > > Why does _top_ show the 7906M number? I suspect that it's deeper than > just synth. > > What else can I share to help diagnose this?Let me try this: I can reduce kern.maxswzone to a number lower than 32G. That won't _solve_ the underlying problem, but we can see if it impacts both top and synth. I just did portsnap fetch update and it came up with 300 patches, so that should give us a good run! [snip] -- Don Wilde **************************************************** * What is the Internet of Things but a system * * of systems including humans? * ****************************************************
On Wednesday, 24 June 2020 at 18:51:04 -0700, Donald Wilde wrote:> On 6/24/20, Greg 'groggy' Lehey <grog at freebsd.org> wrote: >> On Wednesday, 24 June 2020 at 9:36:23 -0700, Donald Wilde wrote: >>> All, >>> >>> I recently upgraded my 12-STABLE system to the latest, and now my >>> swap subsystems aren't working. I deliberately set up a 40GB >>> partition for swap, and when I do 'top -t' I am only seeing 7906M >>> total. >> >> That looks suspiciously like the difference from 32 GB. Could it be >> numeric overflow? And if so, where? What does pstat -s say? > > Well, hi Greg! LTNT2!Indeed.> pstat -shm: > > /dev/ada0s1b 65536 (1M blocks), Used: 1.5G, Avail: 63G, Capacity: 2%Now that's really puzzling. Why does it say 64 G when you said 40 G, and the error from top tends to confirm it? How big is the partition (gpart output)?>>> When using synth, it shows swap usage of < 2%, but then it gives the >>> 'out of swap' complaint and tosses all my users out into neverland. >> >> Unless this also causes kernel messages to that effect, this, too, >> could be an issue with synth. > > I believe the messages about 'out of swap' were from the kernel as > it died, not synth. As I say, it killed all my active users, both > root and non-root.Hmm.> Why does _top_ show the 7906M number? I suspect that it's deeper > than just synth.If top counts swap in a form that overflows at 32 GB, 40 GB could show up as 8 GB.> What else can I share to help diagnose this?Background, maybe? You say that you upgraded your system. Did you change the swap size when you did? What were swap and RAM sizes before and after? FWIW, I've had a very large number of issues with -STABLE in recent months, though not this one. I should send a separate message. Greg -- Sent from my desktop computer. Finger grog at FreeBSD.org for PGP public key. See complete headers for address and phone numbers. This message is digitally signed. If your Microsoft mail program reports problems, please read http://lemis.com/broken-MUA -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 163 bytes Desc: not available URL: <http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20200625/08ddd024/attachment.sig>