Requests for -current (14.2-->15.0)

mats_b_tegner · 11-26-2018, 07:44 PM

GKrellM still asks for old libsensors.so.4 library:
gkrellm: error while loading shared libraries: libsensors.so.4: cannot open shared object file: No such file or directory
even if I rebuild it. What could cause this?

Quote:

Mon Nov 26 22:58:11 UTC 2018
ap/lm_sensors-3.5.0-x86_64-1.txz: Upgraded.
Shared library .so-version bump.
xap/gkrellm-2.3.10-x86_64-5.txz: Rebuilt.
Recompiled against lm_sensors-3.5.0.

Rebooting doesn't solve it for me...found it I had a copy in /usr/local/bin...

cwizardone · 11-26-2018, 07:58 PM

Working fine here after today's updates. I did reboot after installing the updates, but don't know if that had anything at all to do with it.

willysr · 11-26-2018, 08:00 PM

working fine here without reboot

dchmelik · 11-26-2018, 09:15 PM

Quote:

Originally Posted by Alien Bob

I do not have hundreds of files open in a text editor but I do have multiple tens of tabs open in my browser, and I have many Konsole terminals open.

I have 100+ browser tabs open and maybe TB of stuff open in Transmission and two similar programs. I tried KDE5 on a different distribution and it does seem to be improving now even when indexing TB of data. Yes, I was going to retry on Slackware-current, but when I followed the instructions, and uncommented the Slackware-current mirror finder, slackpkg didn't upgrade it, then I needed to temporarily use a different distribution for AMD Radeon RX Vega display/video/graphics card GPU compute support, though have other computers on my desk (laptop, servers) running Slackware.

Quote:

Do you use proprietary video drivers on that computer (Nvidia or AMD)?

no.

I think one of the problems is notifications in the KDE5 system tray will cause the GUI to do something like a 'memory thrashing' problem for user interface input. Ignore a notification (like IRC message with flashing system tray) too long and then soon the mouse will only move a short distance before pausing, again and again. Only restarting X fixes it.

Some other things I'd like to see in Slackware 15 are below.

support for OpenCL on Vega cards (needs AMDGPU-PRO)

all the good 'light' window managers, like (maybe mentioned one or two)

CDE (my favourite, if it has multi-monitor setup or even scroll wheel, not sure, but maybe not really ready for general modern usage)
Enlightenment (what I'd actually use if I could configure monitors and if it had clipboard)
FVWM(95)?
one or two WMs names that start with I? (ICEWM? I3WM? heard of them)
LXDE? (never tried but heard it's faster than XFCE now)
TDE (KDE3.5 fork)

birdboy · 11-27-2018, 12:00 PM

Like Sendmail/Postfix, Courier/Dovecot replacements etc, how about replacing ImageMagick with GraphicsMagick, a much improved fork?

Also, BIND -> nsd/unbound (and maybe add ldns/drill too, as dig substitute).

ntpd is another contender that is long in the tooth. There's NTPsec, but IMO, chrony or OpenNTPD would be better. Ntimed is stalled. chrony is Linux native and looks like the best all around choice.

atelszewski · 11-27-2018, 12:08 PM

Hi,

I'm all for chrony.

--
Best regards,
Andrzej Telszewski

SCerovec · 11-27-2018, 01:16 PM

..., and provide symlinks for the transition period?

bassmadrigal · 11-27-2018, 11:13 PM

I'd like to suggest a minor update to rc.nfsd that was suggested to a user by upnort here 9ish months ago.

I was looking at my dmesg starting up and I was getting an error with nfsd saying: "Unable to end grace period: -110". I had no problems with my nfs shares and there seemed to be no ill effects from this error other than dirtying my dmesg output

Essentially, anytime I would start nfsd, I would get the following output:

Code:

[ 7757.071312] NFSD: the nfsdcld client tracking upcall will be removed in 3.10. Please transition to using nfsdcltrack.
[ 7757.071314] NFSD: starting 45-second grace period (net f0000098)
[ 7834.070500] NFSD: Unable to end grace period: -110

There were a bunch of random unrelated suggestions on google, but then I came across upnort's suggestion to add the following to the beginning of the nfsd_start function in rc.nfsd and it got rid of the error message.

Code:

  # Without this directory the logs will complain with
  # 'NFSD: Unable to end grace period'.
  NFSV4RECOVERYDIR=$(cat /proc/fs/nfsd/nfsv4recoverydir)
  if [ -r /proc/fs/nfsd/nfsv4recoverydir ] && [ ! -d "$NFSV4RECOVERYDIR" ]; then
    mkdir "$NFSV4RECOVERYDIR"
    chown -R rpc:rpc "$NFSV4RECOVERYDIR"
  fi

Adding this also gets rid of the nfscld client error (top line above) as well, and now my output when starting is simply:

Code:

[ 7890.376350] NFSD: Using /var/lib/nfs/v4recovery as the NFSv4 state recovery directory
[ 7890.376359] NFSD: starting 45-second grace period (net f0000098)

Here's a patch:

Code:

diff --git a/rc.nfsd b/rc.nfsd
index 965db6b..d158e57 100644
--- a/rc.nfsd
+++ b/rc.nfsd
@@ -16,6 +16,14 @@ if [ -r /etc/default/nfs ]; then
 fi
 
 nfsd_start() {
+  # Without this directory the logs will complain with
+  # 'NFSD: Unable to end grace period'.
+  NFSV4RECOVERYDIR=$(cat /proc/fs/nfsd/nfsv4recoverydir)
+  if [ -r /proc/fs/nfsd/nfsv4recoverydir ] && [ ! -d "$NFSV4RECOVERYDIR" ]; then
+    mkdir "$NFSV4RECOVERYDIR"
+    chown -R rpc:rpc "$NFSV4RECOVERYDIR"
+  fi
+
   # Sanity checks.  Exit if there's no /etc/exports, or if there aren't any
   # shares defined in it.
   if [ ! -r /etc/exports ]; then # no config file, exit:

gmgf · 11-28-2018, 02:48 AM

samba-4.9.3:

(many CVE fix)
https://www.samba.org/samba/history/
https://download.samba.org/pub/samba/samba-4.9.3.tar.gz

Didier Spaier · 11-28-2018, 03:00 AM

Thank your Pat for allowing usage of F2FS to install Slackware current on a flash drive. Although I personally wouldn't advise to do that (I'd rather use a small USB HD or SSD), I know that there are many users wanting to carry Slackware on the go on an USB stick.

GazL · 11-28-2018, 04:55 AM

May I suggest:

Code:

--- a/rc.S      2018-11-28 10:38:34.015032493 +0000
+++ b/rc.S      2018-11-28 10:48:03.752882922 +0000
@@ -231,7 +231,7 @@
 
 # If we're using F2FS for the root filesystem, don't check it as it doesn't
 # allow checking a read-only filesystem:
-if [ "$(cat /etc/fstab | tr '\t' ' ' | tr -s ' ' | grep ' / ' | cut -f 3 -d ' ')" = "f2fs" ]; then
+if grep -q '^[^#][^[:space:]]*[[:space:]]\+/[[:space:]]\+f2fs[[:space:]]' /etc/fstab ; then
   echo "Remounting root device with read-write enabled."
   /sbin/mount -w -v -n -o remount /
 elif [ ! $READWRITE = yes ]; then

Setting aside the UUOC and all the unnecessary utility calls, yours doesn't ignore commented out lines.

P.S.
Thinking about it a little more, if you used /proc/mounts, instead of /etc/fstab you wouldn't need to worry about the commented lines or repeated spacing, so a '^[^[:space:]]\+ / f2fs ' would suffice.

dugan · 11-28-2018, 08:49 AM

I see that vulkan-sdk 1.1.92.1 is officially out.

bassmadrigal · 11-28-2018, 10:09 AM

Quote:

Originally Posted by Didier Spaier

Although I personally wouldn't advise to do that (I'd rather use a small USB HD or SSD)

This isn't just for USB sticks, it is for NAND-based flash devices, which includes SSDs and NVMes.

mats_b_tegner · 11-28-2018, 11:30 AM

https://blog.kitware.com/cmake-3-13-...-for-download/
https://github.com/Kitware/CMake/rel...-3.13.1.tar.gz

upnort · 11-28-2018, 11:53 AM

Quote:

I'd like to suggest a minor update to rc.nfsd that was suggested to a user by upnort here 9ish months ago.

Thank you for the reference.

There is another quirk with NFS. The /var/lib/rmtab file will accumulate stale NFS connections. The man page states that using this file is not dependable. I don't why the file can't be maintained in real-time by NFS, but c'est la vie.

I don't know the rc.nfsd script is an appropriate place for a fix. Probably rc.local or rc.local_shutdown is more appropriate.

Code:

# Clean NFS logs.
cat /dev/null > /var/lib/nfs/rmtab 2>/dev/null

Then again, no argument from me if that snippet was in the rc.nfsd script.

While not resolving the file being undependable, the scrubbing at least keeps the file reasonably current for the session. Probably does not help much with servers that are on 24/7.

Of course, the biggest quirk with NFS is how clients utterly freeze when the NFS server disappears.