Last night after upgrading several Fedora 27 systems with the latest
http://packages.x2go.org/fedora/$releasever/main/$basearch
nxagent now segfaults sometimes on connection, sometimes when doing a move of the x2goclient screen.
Problem is, I can't roll back to the previous version (to test) as the repo does not contain the older rpms?
Would it be possible to have a version or two to be able to do a "dnf downgrade"?
Cheers.
-- Norman Gaywood, Computer Systems Officer School of Science and Technology University of New England Armidale NSW 2351, Australia
ngaywood@une.edu.au http://turing.une.edu.au/~ngaywood Phone: +61 (0)2 6773 2412 Mobile: +61 (0)4 7862 0062
Please avoid sending me Word or Power Point attachments. See http://www.gnu.org/philosophy/no-word-attachments.html
- On 03/13/2018 07:34 AM, Norman Gaywood wrote:
Last night after upgrading several Fedora 27 systems with the latest
http://packages.x2go.org/fedora/$releasever/main/$basearch
nxagent now segfaults sometimes on connection, sometimes when doing a move of the x2goclient screen.
Problem is, I can't roll back to the previous version (to test) as the repo does not contain the older rpms?
Would it be possible to have a version or two to be able to do a "dnf downgrade"?
You can switch to old versions using http://snapshots.packages.x2go.org/, but these won't be available for a very long time. The retention policy is somewhat complicated.
Even better would be to fix that problem, though...
Mihai
Thanks Mihai,
But all those links/folders on http://snapshots.packages.x2go.org/ seem to be empty.
Or am I missing something?
Fixing the bug could take weeks, so it would be good to roll back 1 version that seemed to be more stable while we wait.
Cheers.
On Tue, 13 Mar 2018 at 20:03, Mihai Moldovan <ionic@ionic.de> wrote:
- On 03/13/2018 07:34 AM, Norman Gaywood wrote:
Last night after upgrading several Fedora 27 systems with the latest
http://packages.x2go.org/fedora/$releasever/main/$basearch
nxagent now segfaults sometimes on connection, sometimes when doing a move of the x2goclient screen.
Problem is, I can't roll back to the previous version (to test) as the repo does not contain the older rpms?
Would it be possible to have a version or two to be able to do a "dnf downgrade"?
You can switch to old versions using http://snapshots.packages.x2go.org/, but these won't be available for a very long time. The retention policy is somewhat complicated.
Even better would be to fix that problem, though...
Mihai
-- Norman Gaywood, Computer Systems Officer School of Science and Technology University of New England Armidale NSW 2351, Australia
ngaywood@une.edu.au http://turing.une.edu.au/~ngaywood Phone: +61 (0)2 6773 2412 Mobile: +61 (0)4 7862 0062
Please avoid sending me Word or Power Point attachments. See http://www.gnu.org/philosophy/no-word-attachments.html
- On 03/13/2018 10:25 AM, Norman Gaywood wrote:
But all those links/folders on http://snapshots.packages.x2go.org/ seem to be empty.
Or am I missing something?
Nope, they are not. See for instance https://snapshots.packages.x2go.org/20180302/fedora/27/main/x86_64/rpms/nx-l...
It's just the index page that is "blank", just like https://packages.x2go.org/ is.
Fixing the bug could take weeks, so it would be good to roll back 1 version that seemed to be more stable while we wait.
The other version has a different severe bug, related to Screen Savers and DPMS. Make sure to disable them, if you really want to downgrade.
Mihai
Ah yes, thanks.
Now to do some more testing :-(
On Tue, 13 Mar 2018 at 20:29, Mihai Moldovan <ionic@ionic.de> wrote:
- On 03/13/2018 10:25 AM, Norman Gaywood wrote:
But all those links/folders on http://snapshots.packages.x2go.org/ seem to be empty.
Or am I missing something?
Nope, they are not. See for instance
https://snapshots.packages.x2go.org/20180302/fedora/27/main/x86_64/rpms/nx-l...
It's just the index page that is "blank", just like https://packages.x2go.org/ is.
Fixing the bug could take weeks, so it would be good to roll back 1 version that seemed to be more stable while we wait.
The other version has a different severe bug, related to Screen Savers and DPMS. Make sure to disable them, if you really want to downgrade.
Mihai
-- Norman Gaywood, Computer Systems Officer School of Science and Technology University of New England Armidale NSW 2351, Australia
ngaywood@une.edu.au http://turing.une.edu.au/~ngaywood Phone: +61 (0)2 6773 2412 Mobile: +61 (0)4 7862 0062
Please avoid sending me Word or Power Point attachments. See http://www.gnu.org/philosophy/no-word-attachments.html
On Tue, 13 Mar 2018 at 20:29, Mihai Moldovan <ionic@ionic.de> wrote:
- On 03/13/2018 10:25 AM, Norman Gaywood wrote:
https://snapshots.packages.x2go.org/20180302/fedora/27/main/x86_64/rpms/nx-l...
Fixing the bug could take weeks, so it would be good to roll back 1 version that seemed to be more stable while we wait.
The other version has a different severe bug, related to Screen Savers and DPMS. Make sure to disable them, if you really want to downgrade.
Thanks for those snapshot links. I manage to use them an roll back 1 version on the dev system to avoid:
https://github.com/ArcticaProject/nx-libs/issues/679 (Coredump when connecting) https://github.com/ArcticaProject/nx-libs/issues/677 (nxagent crashes when moving the session window into another screen)
I was about to roll back production, but users that were getting seg faults (on connecting or moving across screens) seemed to start working!
So I've left production alone for now.
Of the 5/6 users that hit these problems most of them seem to be connecting again.
Seg faults are still occurring in nxagent and some users still have not connected successfully, but most seem fine and get on after a few retries.
???
3.5.99.14-0.0build1.1.git20180302.2954.main.fc27 (dev) 3.5.99.15-0.0build1.1.git20180308.2971.main.fc27 (prod)
-- Norman Gaywood, Computer Systems Officer School of Science and Technology University of New England Armidale NSW 2351, Australia
ngaywood@une.edu.au http://turing.une.edu.au/~ngaywood Phone: +61 (0)2 6773 2412 Mobile: +61 (0)4 7862 0062
Please avoid sending me Word or Power Point attachments. See http://www.gnu.org/philosophy/no-word-attachments.html
So I've updated dev to the latest again 3.5.99.15-0.0build1.1.git20180308.2971.main.fc27
And I seem to get an nxagent segfault every time I move the session window into another screen.
Same version in production and it was stable for a few days now. I could move it across screens. I suspended and resumed several times. But, after logging out of production and logging back in again, I'm back to segfaulting when moving across screens.
It's a very strange problem :-(
Executive summary:
The segfault on connect I have not reproduced. But its been reported by several of my users. I have around 50 logged in users at the moment. The login segfault only happens to very few (5/6). Some of them seemed to be working now.
The move across the screen segfault has happened to 3 users (including me) and all have been linux clients (Fedora I think). The move across screen segfault seems to go away sometimes for undermined reasons.
On Thu, 15 Mar 2018 at 12:36, Norman Gaywood <ngaywood@une.edu.au> wrote:
On Tue, 13 Mar 2018 at 20:29, Mihai Moldovan <ionic@ionic.de> wrote:
- On 03/13/2018 10:25 AM, Norman Gaywood wrote:
https://snapshots.packages.x2go.org/20180302/fedora/27/main/x86_64/rpms/nx-l...
Fixing the bug could take weeks, so it would be good to roll back 1 version that seemed to be more stable while we wait.
The other version has a different severe bug, related to Screen Savers and DPMS. Make sure to disable them, if you really want to downgrade.
Thanks for those snapshot links. I manage to use them an roll back 1 version on the dev system to avoid:
https://github.com/ArcticaProject/nx-libs/issues/679 (Coredump when connecting) https://github.com/ArcticaProject/nx-libs/issues/677 (nxagent crashes when moving the session window into another screen)
I was about to roll back production, but users that were getting seg faults (on connecting or moving across screens) seemed to start working!
So I've left production alone for now.
Of the 5/6 users that hit these problems most of them seem to be connecting again.
Seg faults are still occurring in nxagent and some users still have not connected successfully, but most seem fine and get on after a few retries.
???
3.5.99.14-0.0build1.1.git20180302.2954.main.fc27 (dev) 3.5.99.15-0.0build1.1.git20180308.2971.main.fc27 (prod)
-- Norman Gaywood, Computer Systems Officer School of Science and Technology University of New England Armidale NSW 2351, Australia
ngaywood@une.edu.au http://turing.une.edu.au/~ngaywood Phone: +61 (0)2 6773 2412 Mobile: +61 (0)4 7862 0062
Please avoid sending me Word or Power Point attachments. See http://www.gnu.org/philosophy/no-word-attachments.html
-- Norman Gaywood, Computer Systems Officer School of Science and Technology University of New England Armidale NSW 2351, Australia
ngaywood@une.edu.au http://turing.une.edu.au/~ngaywood Phone: +61 (0)2 6773 2412 Mobile: +61 (0)4 7862 0062
Please avoid sending me Word or Power Point attachments. See http://www.gnu.org/philosophy/no-word-attachments.html
- On 03/15/2018 03:11 AM, Norman Gaywood wrote:
So I've updated dev to the latest again 3.5.99.15-0.0build1.1.git20180308.2971.main.fc27
And I seem to get an nxagent segfault every time I move the session window into another screen.
Same version in production and it was stable for a few days now. I could move it across screens. I suspended and resumed several times. But, after logging out of production and logging back in again, I'm back to segfaulting when moving across screens.
It's a very strange problem :-(
Executive summary:
The segfault on connect I have not reproduced. But its been reported by several of my users. I have around 50 logged in users at the moment. The login segfault only happens to very few (5/6). Some of them seemed to be working now.
I've been unable to reproduce this on a 16.04 VM with two virtual heads, but neither was the original reporter that tried connecting from a 16.04 with one head to a 16.04 machine, that previously segfaulted on connect when using a 14.04 client box with dual head setup (with 21:9 displays or such).
The move across the screen segfault has happened to 3 users (including me) and all have been linux clients (Fedora I think). The move across screen segfault seems to go away sometimes for undermined reasons.
I personally wasn't able to reproduce that either, but it seems to be depending on quite some factors.
Can you tell me the display configuration (size and positioning) you are seeing this with and what window managers/desktop environments and operating systems are used on the client and server side?
Mihai
On Thu, 15 Mar 2018 at 13:45, Mihai Moldovan <ionic@ionic.de> wrote:
The move across the screen segfault has happened to 3 users (including me) and all have been linux clients (Fedora I think). The move across screen segfault seems to go away sometimes for undermined reasons.
I personally wasn't able to reproduce that either, but it seems to be depending on quite some factors.
If I connect, suspend the session, resume the session. I can then move my session across screens. Worked for me a few times so far.
Can you tell me the display configuration (size and positioning) you are seeing this with and what window managers/desktop environments and operating systems are used on the client and server side?
Not sure I can in a useful manner, but: I keep changing my connection screen size to many different sizes. I'm not sure how to report my physical workstation screen sizes but: Dell 27" 2560x1440, Dell 29" 2560x1018 The workstation x2goclient is x2goclient-4.1.1.0-1.fc27.x86_64
All desktop environments on the server(s) are XFCE, xfce4-session-4.12.1-12.fc27.x86_64 The server(s) are running: libNX_X11-6-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 libXcompext3-3.5.0.33-0.0x2go1.1.git20180204.776.main.fc27.x86_64 nxagent-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nx-libs-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nxproxy-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 perl-X2Go-Log-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-DB-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-common-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-extensions-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-x2goagent-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-xsession-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64
The prod server would have mostly Windows x2go clients connecting, but some Mac and some Linux, all different versions. The dev server would only have x2goclient-4.1.1.0-1.fc27.x86_64 connecting.
The servers (dev and prod) are Fedora 27 running: libNX_X11-6-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 libXcompext3-3.5.0.33-0.0x2go1.1.git20180204.776.main.fc27.x86_64 nxagent-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nx-libs-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nxproxy-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 perl-X2Go-Log-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-DB-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-common-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-extensions-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-x2goagent-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-xsession-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64
-- Norman Gaywood, Computer Systems Officer School of Science and Technology University of New England Armidale NSW 2351, Australia
ngaywood@une.edu.au http://turing.une.edu.au/~ngaywood Phone: +61 (0)2 6773 2412 Mobile: +61 (0)4 7862 0062
Please avoid sending me Word or Power Point attachments. See http://www.gnu.org/philosophy/no-word-attachments.html
- On 03/15/2018 05:23 AM, Norman Gaywood wrote:
If I connect, suspend the session, resume the session. I can then move my session across screens. Worked for me a few times so far. Hmm, I just tried that right now and couldn't. nxagent crashes, even after suspending and resuming the session before moving it to another display.
Not sure I can in a useful manner, but: I keep changing my connection screen size to many different sizes. I'm not sure how to report my physical workstation screen sizes but: Dell 27" 2560x1440, Dell 29" 2560x1018
Okay, with configuration I meant something like that:
| P || S | | 2560x1440 || 2560x1018 | | | -----------
Where P denotes a primary display and S a secondary one. Marking just the primary would also be okay, since there can only be one™.
In this configuration, the secondary display is east of the other one and aligned at the top.
Anyway, I am retracting my initial assessment, I can reproduce the crashing when moving nxagent windows around display, but only iff the display sizes differ.
With two displays of the same size, this doesn't happen.
The workstation x2goclient is x2goclient-4.1.1.0-1.fc27.x86_64
All desktop environments on the server(s) are XFCE, xfce4-session-4.12.1-12.fc27.x86_64 The server(s) are running: libNX_X11-6-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 libXcompext3-3.5.0.33-0.0x2go1.1.git20180204.776.main.fc27.x86_64 nxagent-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nx-libs-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nxproxy-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 perl-X2Go-Log-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-DB-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-common-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-extensions-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-x2goagent-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-xsession-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64
The prod server would have mostly Windows x2go clients connecting, but some Mac and some Linux, all different versions. The dev server would only have x2goclient-4.1.1.0-1.fc27.x86_64 connecting. The servers (dev and prod) are Fedora 27 running: libNX_X11-6-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 libXcompext3-3.5.0.33-0.0x2go1.1.git20180204.776.main.fc27.x86_64 nxagent-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nx-libs-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 nxproxy-3.5.99.15-0.0build1.1.git20180308.2971.main.fc27.x86_64 perl-X2Go-Log-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 perl-X2Go-Server-DB-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-common-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-extensions-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-x2goagent-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64 x2goserver-xsession-4.1.0.0-0.0x2go1.1.git20180302.1517.main.fc27.x86_64
Everything's okay. You have an old version of libXcompext3 installed, but it's an orphaned leaf and won't cause trouble since nothing links against it.
You may remove it manually if you wish.
Mihai
-
Mihai Moldovan -
Norman Gaywood