From Patrick.Fleckenstein at technolit.de Tue Jan 7 11:08:20 2020 From: Patrick.Fleckenstein at technolit.de (Patrick Fleckenstein) Date: Tue, 7 Jan 2020 11:08:20 +0000 Subject: [winswitch] [XPRA] server-side performance with many users Message-ID: <9b4ab8aec703204f978c9fc6eacd72bbf1993956.camel@technolit.de> Hi, I'm running XPRA on Ubuntu headless server. 100GB RAM, 16CPUs, 10Gb network interface. If connecting with a small amount of remote machines performance is great. But in my environment nearly 200 users have to log in at the same time with their own account. Therefor I created a script checking on local startup if a xpra process is running for this user. If not it starts. If many users are connected (~90+) performance gets worse with linux load at 200+. Is there a performance hack for such a setup or a best practise? Each user instance have to run separatly in each users shell because auth with ssh which access to the executed program. Any hints on this or is it simply not possible to run xpra with so much users on one machine? Regards Patrick ---------------------------------------------------------------------- Die Firma TECHNOLIT? GmbH erhebt personenbezogene Daten zur Durchf?hrung der mit Ihnen vereinbarten Leistungen. Wir verarbeiten personenbezogene Daten gem?? den Bestimmungen der DS-GVO. Weitere ausf?hrliche Informationen zum Datenschutz finden Sie auf unserer Website www.technolit.de unter Datenschutz. Auf Wunsch ?bersenden wir diese Information auch in Papierform. TECHNOLIT? GmbH Industriestr. 8, 36137 Gro?enl?der Gesch?ftsf?hrer: Simone Lang, Stephan G?nther, Klaus Sch?nherr Amtsgericht Fulda, HRB 390 Diese E-Mail enth?lt vertrauliche und/oder rechtlich gesch?tzte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrt?mlich erhalten, informieren Sie bitte sofort den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail ist nicht gestattet. From antoine at devloop.org.uk Wed Jan 8 12:10:06 2020 From: antoine at devloop.org.uk (Antoine Martin) Date: Wed, 8 Jan 2020 13:10:06 +0100 Subject: [winswitch] [ANNOUNCE] Xpra 3.0.5: many minor fixes, one critical (same one again) Message-ID: Hi, There are a few fixes in this release, none of them are critical apart from the one affecting MS Windows Python 3 clients with OpenGL enabled when displaying undecorated windows. That one was meant to be fixed in 3.0.4, but the patch was mis-applied. There is no urgency to update if you were not affected by these bugs. Release notes: * fix missing undecorated opengl windows on win32 with GTK3 * fix fake Xinerama errors with unicode monitor names * fix av-sync backport for python2 servers * fix errors when the dbus submodule is not installed * fix RFB server key handling * fix exit code for unsupported sockets * fix tray title with openssh and plink SSH backends * fix crashes on win32 shadow exit * fix 'xpra upgrade' * fix focus problems with OR windows * fix wrong client info shown in 'xpra top' * fix 'xpra top' error that scrambled the output, flickering * fix X11 keyboard query against secondary screens * skip trying to load X11 components on non-X11 platforms * allow F11 through to the browser to make it easier to go fullscreen * prevent sshfp dns errors from causing ssh connection failures * obscure passwords from log files * remove duplicated attribute from xpra info * remove unused loggers * expose av-sync and webcam client capabilities * better wayland mode detection * add more missing files to MANIFEST * add v4l2 codec files to clean target * add logging to gtk display cleanup Cheers, Antoine From ajs1 at cam.ac.uk Thu Jan 9 19:04:00 2020 From: ajs1 at cam.ac.uk (Anthony Stone) Date: Thu, 9 Jan 2020 19:04:00 +0000 Subject: [winswitch] xpra slow update Message-ID: <29c6d536-baf7-0908-e859-52222f132a40@cam.ac.uk> Hello Antoine, Since version 3.0.4 I find that an emacs window on the remote is very slow to update edits. I have tried setting xpra > picture > speed > lowest latency, but that doesn't seem to help. (Should it?) My download speed is 100 Mbit/s, but the upload speed is only 3 Mbit/s; could that account for it? I don't know whether the speed has changed recently. What happens is that I edit characters on a line in a text file in emacs, and the change often isn't reflected on the screen for several seconds. It is updated promptly if I move the cursor to an adjacent line. Otherwise I often assume I've mistyped, and correct it, only to find that it was correct after all. Anthony From antoine at nagafix.co.uk Thu Jan 9 21:01:50 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Thu, 9 Jan 2020 22:01:50 +0100 Subject: [winswitch] xpra slow update In-Reply-To: <29c6d536-baf7-0908-e859-52222f132a40@cam.ac.uk> References: <29c6d536-baf7-0908-e859-52222f132a40@cam.ac.uk> Message-ID: On 09/01/2020 20:04, Anthony Stone via shifter-users wrote: > Hello Antoine, > > Since version 3.0.4 What version did you have installed previously? 3.0.3? Is this both on the client and the server? There is nothing in the 3.0.3 changesets that should be able to cause this sort of problem. Maybe it's a build or environment issue. > I find that an emacs window on the remote is very > slow to update edits. I have tried setting xpra > picture > speed > > lowest latency, but that doesn't seem to help. (Should it?) It helps tune things, but what you're experiencing sounds like a bug. > My download > speed is 100 Mbit/s, but the upload speed is only 3 Mbit/s; could that > account for it? No. 3Mbps is not ideal, but it should be enough for running a text editor for sure. > I don't know whether the speed has changed recently. > > What happens is that I edit characters on a line in a text file in > emacs, and the change often isn't reflected on the screen for several > seconds. It is updated promptly if I move the cursor to an adjacent > line. Otherwise I often assume I've mistyped, and correct it, only to > find that it was correct after all. Please file a ticket with more details, including log files. There must be something broken. Cheers, Antoine > > Anthony > _______________________________________________ > shifter-users mailing list > shifter-users at lists.devloop.org.uk > https://lists.devloop.org.uk/mailman/listinfo/shifter-users > From wuebbel at gmail.com Fri Jan 10 14:02:28 2020 From: wuebbel at gmail.com (wuebbel at gmail.com) Date: Fri, 10 Jan 2020 15:02:28 +0100 Subject: [winswitch] xpra 3.0.5: Message-ID: <012701d5c7be$98ae6fa0$ca0b4ee0$@gmail.com> Dear list, connecting through the HTML client in 3.0.5 is not working for me. Downgrading to 3.0.4, everything is fine. Details: - Ubuntu 18.04.3 - Installed xpra through the winswitch-repositories as proposed in the install section on xpra.org After installing via apt-get install xpra=3.0.4-r24778-1 xpra-html5=3.0.4-r24778-1 and connecting to Port 14500 via web browser, xpra fails at first (socket closed), due to Jan 10 14:41:54 RADON xpra[30678]: Error: failed to setup control socket '/run/xpra/RADON-proxy-31065': Changing the permissions on /run/xpra (or adding the user to the xpra group) fixes this, and everything is fine, I have been running this configuration for a couple of days. However: After doing apt-get upgrade (or purging and installing 3.0.5 for both xpra and xpra.html-5), and again connecting via the webbrowser to port 14500, it fails and shows: Did not receive hello before timeout reached, not an Xpra server? Syslog shows no errors. Input is exactly the same as in the working version. Starting the server manually via xpra start and then doing xpra attach or xpra launcher works, but still, html5 fails. Any ideas? I tried to use debug, but was unsure where to locate this problem. Frank From antoine at nagafix.co.uk Fri Jan 10 18:50:01 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Fri, 10 Jan 2020 19:50:01 +0100 Subject: [winswitch] [XPRA] server-side performance with many users In-Reply-To: <9b4ab8aec703204f978c9fc6eacd72bbf1993956.camel@technolit.de> References: <9b4ab8aec703204f978c9fc6eacd72bbf1993956.camel@technolit.de> Message-ID: On 07/01/2020 12:08, Patrick Fleckenstein via shifter-users wrote: > Hi, > > I'm running XPRA on Ubuntu headless server. 100GB RAM, 16CPUs, 10Gb network interface. > If connecting with a small amount of remote machines performance is great. > > But in my environment nearly 200 users have to log in at the same time with their own account. > Therefor I created a script checking on local startup if a xpra process is running for this user. > If not it starts. > > If many users are connected (~90+) performance gets worse with linux load at 200+. 90 is a quite a lot for just 16CPUs. > Is there a performance hack for such a setup or a best practise? Here are some ideas: 1) turn off any option you don't need, ie: in particular turn off audio if you don't need it and don't bother starting pulseaudio at all 2) sacrifice some bandwidth for CPU usage by increasing the default value for "min-speed" (assuming that you have bandwidth to spare) 3) sacrifice some quality: reduce "min-quality" 4) use picture encodings that are light on CPU usage: rgb+lz4 or jpeg on low quality (this will cost extra bandwidth) - avoid png at all cost 5) tune further to reduce CPU usage - this may require using environment variables which are only documented in the source code 6) tune your hypervisor, reduce context switching etc.. > Each user instance have to run separatly in each users shell because auth with ssh which access to the executed program. > > Any hints on this or is it simply not possible to run xpra with so much users on one machine? It really depends on what your application does, how much CPU and graphics bandwidth it requires. 90 users is quite high. Cheers, Antoine > > Regards > Patrick > > > ---------------------------------------------------------------------- > Die Firma TECHNOLIT? GmbH erhebt personenbezogene Daten zur Durchf?hrung der mit Ihnen vereinbarten Leistungen. Wir verarbeiten personenbezogene Daten gem?? den Bestimmungen der DS-GVO. > Weitere ausf?hrliche Informationen zum Datenschutz finden Sie auf unserer Website www.technolit.de unter Datenschutz. > Auf Wunsch ?bersenden wir diese Information auch in Papierform. > > TECHNOLIT? GmbH > Industriestr. 8, 36137 Gro?enl?der > Gesch?ftsf?hrer: Simone Lang, Stephan G?nther, Klaus Sch?nherr > Amtsgericht Fulda, HRB 390 > > Diese E-Mail enth?lt vertrauliche und/oder rechtlich gesch?tzte > Informationen. Wenn Sie nicht der richtige Adressat sind oder diese > E-Mail irrt?mlich erhalten, informieren Sie bitte sofort den Absender > und vernichten Sie diese Mail. Das unerlaubte Kopieren sowie > die unbefugte Weitergabe dieser Mail ist nicht gestattet. > > > > _______________________________________________ > shifter-users mailing list > shifter-users at lists.devloop.org.uk > https://lists.devloop.org.uk/mailman/listinfo/shifter-users > From antoine at nagafix.co.uk Fri Jan 10 22:02:08 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Fri, 10 Jan 2020 23:02:08 +0100 Subject: [winswitch] xpra 3.0.5: In-Reply-To: <012701d5c7be$98ae6fa0$ca0b4ee0$@gmail.com> References: <012701d5c7be$98ae6fa0$ca0b4ee0$@gmail.com> Message-ID: <8d6f243b-e6f2-0c79-66d2-f3e13a949b1d@nagafix.co.uk> On 10/01/2020 15:02, wuebbel--- via shifter-users wrote: > Dear list, > > connecting through the HTML client in 3.0.5 is not working for me. > Downgrading to 3.0.4, everything is fine. TLDR: just run apt / dnf / yum update again and things should be fixed. > Details: > > - Ubuntu 18.04.3 > - Installed xpra through the winswitch-repositories as proposed in the > install section on xpra.org > > After installing via > apt-get install xpra=3.0.4-r24778-1 xpra-html5=3.0.4-r24778-1 > and connecting to Port 14500 via web browser, xpra fails at first (socket > closed), due to > Jan 10 14:41:54 RADON xpra[30678]: Error: failed to setup control socket > '/run/xpra/RADON-proxy-31065': > > Changing the permissions on /run/xpra (or adding the user to the xpra group) > fixes this, and everything is fine, I have been running this configuration > for a couple of days. > > However: After doing > apt-get upgrade > (or purging and installing 3.0.5 for both xpra and xpra.html-5), and again > connecting via the webbrowser to port 14500, it fails and shows: > > Did not receive hello before timeout reached, not an Xpra server? > > Syslog shows no errors. Input is exactly the same as in the working version. > > Starting the server manually via xpra start and then doing xpra attach or > xpra launcher works, but still, html5 fails. > > Any ideas? I tried to use debug, but was unsure where to locate this > problem. This is probably an error in the html5 client code, which you can see in your browser's Javascript console. I believe you need this bug fix (two commits because PEBKAC): https://xpra.org/trac/changeset/24934 https://xpra.org/trac/changeset/24939 I have uploaded new 3.0.5 packages with this simple fix applied. Updating your system should give you the fixed packages. You can also use 3.0.5 with version "3.0.4" of the "xpra-html5" package. Sorry about that. Antoine > > Frank From igor at mir2.org Wed Jan 15 13:56:31 2020 From: igor at mir2.org (Igor Bukanov) Date: Wed, 15 Jan 2020 14:56:31 +0100 Subject: [winswitch] UDP connection issues Message-ID: Hi, I am trying to setup a udp connection to xpra server on Linux from a a Linux client. Both server and client uses 3.0.5 version. I run the server using xpra start --bind-udp=0.0.0.0:port --udp-auth=file:filename=.config/xpra/password :28 I can connect to it using ssh, but when I tried on the client: xpra attach udp://server-ip:port/28 The client reported an error: 2020-01-15 14:28:23,228 Xpra GTK3 X11 client version 3.0.5-r24939 64-bit ... 2020-01-15 14:28:25,966 Error: failed to connect, received 2020-01-15 14:28:25,966 '8000000032108494958117100112459911111011611411110810548101105481011054810110548101100101108101101' On the server side I see from the log: 2020-01-15 13:30:25,696 Error: connection timed out: udp socket: 0.0.0.0:server_port <- client_ip:52119 The connection is done over OpenVPN and I can ping the client ip from the server. Any clues? From antoine at nagafix.co.uk Sun Jan 19 14:18:14 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Sun, 19 Jan 2020 21:18:14 +0700 Subject: [winswitch] UDP connection issues In-Reply-To: References: Message-ID: <4586e164-6a54-915d-7b37-77aced3bc5ae@nagafix.co.uk> Sorry for the slow response, this was a tricky one. On 15/01/2020 20:56, Igor Bukanov via shifter-users wrote: > Hi, > > I am trying to setup a udp connection to xpra server on Linux from a a > Linux client. Both server and client uses 3.0.5 version. I run the > server using > > xpra start --bind-udp=0.0.0.0:port > --udp-auth=file:filename=.config/xpra/password :28 > > I can connect to it using ssh, but when I tried on the client: > > xpra attach udp://server-ip:port/28 That's all correct. > The client reported an error: > > 2020-01-15 14:28:23,228 Xpra GTK3 X11 client version 3.0.5-r24939 64-bit > ... > 2020-01-15 14:28:25,966 Error: failed to connect, received > 2020-01-15 14:28:25,966 > '8000000032108494958117100112459911111011611411110810548101105481011054810110548101100101108101101' > > On the server side I see from the log: > > 2020-01-15 13:30:25,696 Error: connection timed out: udp socket: > 0.0.0.0:server_port <- client_ip:52119 > > The connection is done over OpenVPN and I can ping the client ip from > the server. Note that UDP has not been tested or used much lately and there will be some improvements in the next release as I've just spotted some sub-optimal behaviour. > > Any clues? Yes: there was a bug affecting all python3 builds, and since v3 switched to python3 by default, this broke UDP for default installations. (we should add regression tests for this) This is now fixed and will be included in the next stable update. For the gory details, see: https://xpra.org/trac/ticket/2554 Cheers, Antoine From mcouteau at twc.com Tue Jan 21 00:55:57 2020 From: mcouteau at twc.com (Mark Couteau) Date: Mon, 20 Jan 2020 18:55:57 -0600 Subject: [winswitch] xpra-3.0.5 header file issue Message-ID: <8fa74fc3-795a-8021-d3f4-b99ca052d664@twc.com> Hi List, I was working to get xpra-3.0.5 to compile on Linux Mint. I got this error from running setup.py /usr/include/gtk-2.0/gtk/gtkitemfactory.h:47:1: error: function declaration isn?t a prototype [-Werror=strict-prototypes] ?typedef void (*GtkItemFactoryCallback)? (); ?????????????? ^~~~~~~ I commented out the GtkItemFactoryCallback definition in the header file and changed the reference to it to the GtkItemFactoryCallback1 definition.? The program did compile and works so far. I don't know if this is a real issue or something that only caught me up.? I had this same issue on both Intel and AMD machines. Thanks From antoine at nagafix.co.uk Tue Jan 21 05:55:48 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Tue, 21 Jan 2020 12:55:48 +0700 Subject: [winswitch] xpra-3.0.5 header file issue In-Reply-To: <8fa74fc3-795a-8021-d3f4-b99ca052d664@twc.com> References: <8fa74fc3-795a-8021-d3f4-b99ca052d664@twc.com> Message-ID: <3675100f-a3a0-7a44-a1d2-cb39b11e650f@nagafix.co.uk> On 21/01/2020 07:55, Mark Couteau via shifter-users wrote: > Hi List, > > I was working to get xpra-3.0.5 to compile on Linux Mint. This must be an old version. > I got this error from running setup.py > > /usr/include/gtk-2.0/gtk/gtkitemfactory.h:47:1: error: function > declaration isn?t a prototype [-Werror=strict-prototypes] > ?typedef void (*GtkItemFactoryCallback)? (); > > ?????????????? ^~~~~~~ > > I commented out the GtkItemFactoryCallback definition in the header file > and changed the reference to it to the > GtkItemFactoryCallback1 definition.? The program did compile and works > so far. You can undo those changes. You should not be messing with system headers to build things. > I don't know if this is a real issue or something that only caught me > up.? I had this same issue on both Intel and AMD machines. The setup.py file already knows about this issue and will de-activate the warning that causes the build to fail when it detects an Ubuntu distribution that ships with this problematic header file. But your Linux Mint distribution string is not known about, so this doesn't happen automatically. Instead, you can run the build with: python setup.py --without-strict Which will turn off all the warnings (and not just this one) and allow the build to proceed. Cheers, Antoine > > Thanks > > > _______________________________________________ > shifter-users mailing list > shifter-users at lists.devloop.org.uk > https://lists.devloop.org.uk/mailman/listinfo/shifter-users From zhuojia.dai at gmail.com Wed Jan 22 03:05:13 2020 From: zhuojia.dai at gmail.com (Zhuo Jia Dai) Date: Wed, 22 Jan 2020 14:05:13 +1100 Subject: [winswitch] Possible to start HTML showing just one app? Message-ID: Hi all, When I do ``` xpra start --bind-tcp=0.0.0.0:14500 --html=on --start-child=atom ``` I see that `localhost:14500` shows me a desktop-like instead of just the `atom`. Is there a way to start the html client os that only one app is shown and not the desktop? -- ZJ zhuojia.dai at gmail.com From antoine at nagafix.co.uk Wed Jan 22 06:55:48 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Wed, 22 Jan 2020 13:55:48 +0700 Subject: [winswitch] xpra-3.0.5 header file issue In-Reply-To: <107c3c9f-7ffc-7284-d034-2e12b69ebeb1@twc.com> References: <8fa74fc3-795a-8021-d3f4-b99ca052d664@twc.com> <3675100f-a3a0-7a44-a1d2-cb39b11e650f@nagafix.co.uk> <107c3c9f-7ffc-7284-d034-2e12b69ebeb1@twc.com> Message-ID: Please always keep the list CCed. On 22/01/2020 09:31, Mark Couteau wrote: > Hi Antoine, > > The Linux Mint distro version of xpra is this: > xpra/bionic,now 2.1.3+dfsg-1ubuntu1 amd64 [installed] That's dangerously out of date. > so, that one is much older than the 3.0.5 version.? I was unsure being > new to the software if the html version was the way to go. It really depends on your use-case. The html5 client is unlikely to ever reach the same level of performance as the regular client, but on the other hand it doesn't require any packages to be installed apart from a browser. > I have used > vnc and rdp before, but not xpra.? From the tar files out there, html is > getting the latest efforts.? I will try it soon. IIRC, there were many improvements to the html5 client during the 2.4.x and 2.5.x release cycles, not so much in v3. Cheers Antoine > > Thanks > > for the support & guidance > > Mark > > On 1/20/20 11:55 PM, Antoine Martin via shifter-users wrote: >> On 21/01/2020 07:55, Mark Couteau via shifter-users wrote: >>> Hi List, >>> >>> I was working to get xpra-3.0.5 to compile on Linux Mint. >> This must be an old version. >> >>> I got this error from running setup.py >>> >>> /usr/include/gtk-2.0/gtk/gtkitemfactory.h:47:1: error: function >>> declaration isn?t a prototype [-Werror=strict-prototypes] >>> ??typedef void (*GtkItemFactoryCallback)? (); >>> >>> ??????????????? ^~~~~~~ >>> >>> I commented out the GtkItemFactoryCallback definition in the header file >>> and changed the reference to it to the >>> GtkItemFactoryCallback1 definition.? The program did compile and works >>> so far. >> You can undo those changes. You should not be messing with system >> headers to build things. >> >>> I don't know if this is a real issue or something that only caught me >>> up.? I had this same issue on both Intel and AMD machines. >> The setup.py file already knows about this issue and will de-activate >> the warning that causes the build to fail when it detects an Ubuntu >> distribution that ships with this problematic header file. >> But your Linux Mint distribution string is not known about, so this >> doesn't happen automatically. Instead, you can run the build with: >> python setup.py --without-strict >> Which will turn off all the warnings (and not just this one) and allow >> the build to proceed. >> >> Cheers, >> Antoine >> >>> Thanks >>> >>> >>> _______________________________________________ >>> shifter-users mailing list >>> shifter-users at lists.devloop.org.uk >>> https://lists.devloop.org.uk/mailman/listinfo/shifter-users >> _______________________________________________ >> shifter-users mailing list >> shifter-users at lists.devloop.org.uk >> https://lists.devloop.org.uk/mailman/listinfo/shifter-users From antoine at nagafix.co.uk Wed Jan 22 07:02:36 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Wed, 22 Jan 2020 14:02:36 +0700 Subject: [winswitch] Possible to start HTML showing just one app? In-Reply-To: References: Message-ID: On 22/01/2020 10:05, Zhuo Jia Dai via shifter-users wrote: > Hi all, > > When I do > > ``` > xpra start --bind-tcp=0.0.0.0:14500 --html=on --start-child=atom > ``` > > I see that `localhost:14500` shows me a desktop-like instead of just the > `atom`. Is there a way to start the html client os that only one app is > shown and not the desktop? No. Whatever you do, the browser window area is what it is. We can't resize the browser window to match the window size of "atom" (or any other client applications), and we can't have each application window a browser popup either. Now, it's not as bad as it sounds because the "desktop-like" area that you see with xpra is actually fully managed client-side by the browser, unlike full desktops like VNC where everything, including the desktop background, is managed server side and forwarded as pixels. This makes a huge difference when some latency is involved: moving windows is a client-side operation with xpra, server-side with VNC. Even with sub-100ms latency, the usability and bandwidth savings are substantial. Cheers, Antoine From Eric.Lemings at ngc.com Tue Jan 28 22:27:21 2020 From: Eric.Lemings at ngc.com (Lemings, Brad [US] (MS)) Date: Tue, 28 Jan 2020 22:27:21 +0000 Subject: [winswitch] Xpra 3 Test Failures Message-ID: <0c6f0934b76e4940a8d3da2b44ed93ed@XCGC3029.northgrum.com> Hello, I'm trying to rebuild the SRPM downloaded from https://xpra.org/dists/RedHat/7.7/SRPMS/xpra-3.0.5-0.r24939xpra1.el7_7.src.rpm. I get the following test failures. ... 2243 Ran 2 tests in 2.249s 2244 2245 OK 2246 Python Cryptography: 2247 Encryption: 2248 1KB: 0.2ms: 5MB/s 2249 1024KB: 3.2ms: 311MB/s 2250 Decryption: 2251 1KB: 0.2ms: 4MB/s 2252 1024KB: 55.1ms: 18MB/s 2253 Overall: 2254 1KB: 0.1ms: 10MB/s 2255 1024KB: 29.8ms: 33MB/s 2256 running unit.server.server_sockets_test 2257 2258 ..FE. 2259 ====================================================================== 2260 ERROR: test_tcp_socket (__main__.ServerSocketsTest) 2261 ---------------------------------------------------------------------- 2262 Traceback (most recent call last): 2263 File "/rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 71, in test_tcp_socket 2264 self._test_connect(["--bind-tcp=0.0.0.0:%i" % port], "allow", [], b"hello", "ws://127.0.0.1:%i/" % port, EXIT_OK) 2265 File " /rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 47, in _test_connect 2266 raise Exception("version client failed to connect, returned %s" % estr(r)) 2267 Exception: version client failed to connect, returned 1 : CONNECTION_LOST 2268 2269 ====================================================================== 2270 FAIL: test_ssl (__main__.ServerSocketsTest) 2271 ---------------------------------------------------------------------- 2272 Traceback (most recent call last): 2273 File "/rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 143, in test_ssl 2274 test_connect("wss://127.0.0.1:%i/" % tcp_port, EXIT_OK, noverify) 2275 File "/rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 136, in test_connect 2276 assert r==exit_code, "expected info client to return %s but got %s" % (exit_code, client.poll()) 2277 AssertionError: expected info client to return 0 but got None 2278 2279 ---------------------------------------------------------------------- 2280 Ran 5 tests in 81.019s 2281 2282 FAILED (failures=1, errors=1) 2283 failure on unit.server.server_sockets_test, exit code=1 2284 error: Bad exit status from /var/tmp/rpm-tmp.oV8cjR (%check) 2285 2286 2287 RPM build errors: 2288 Bad exit status from /var/tmp/rpm-tmp.oV8cjR (%check) Any clues what might be causing this? Thanks, Eric. From antoine at nagafix.co.uk Thu Jan 30 04:44:10 2020 From: antoine at nagafix.co.uk (Antoine Martin) Date: Thu, 30 Jan 2020 11:44:10 +0700 Subject: [winswitch] Xpra 3 Test Failures In-Reply-To: <0c6f0934b76e4940a8d3da2b44ed93ed@XCGC3029.northgrum.com> References: <0c6f0934b76e4940a8d3da2b44ed93ed@XCGC3029.northgrum.com> Message-ID: <9c7e64d3-0134-7bf3-e4c2-b8ffe1a004f8@nagafix.co.uk> On 29/01/2020 05:27, Lemings, Brad [US] (MS) via shifter-users wrote: > Hello, > > I'm trying to rebuild the SRPM downloaded from https://xpra.org/dists/RedHat/7.7/SRPMS/xpra-3.0.5-0.r24939xpra1.el7_7.src.rpm. > > I get the following test failures. (..) > 2259 ====================================================================== > 2260 ERROR: test_tcp_socket (__main__.ServerSocketsTest) > 2261 ---------------------------------------------------------------------- > 2262 Traceback (most recent call last): > 2263 File "/rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 71, in test_tcp_socket > 2264 self._test_connect(["--bind-tcp=0.0.0.0:%i" % port], "allow", [], b"hello", "ws://127.0.0.1:%i/" % port, EXIT_OK) > 2265 File " /rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 47, in _test_connect > 2266 raise Exception("version client failed to connect, returned %s" % estr(r)) > 2267 Exception: version client failed to connect, returned 1 : CONNECTION_LOST > 2268 > 2269 ====================================================================== > 2270 FAIL: test_ssl (__main__.ServerSocketsTest) > 2271 ---------------------------------------------------------------------- > 2272 Traceback (most recent call last): > 2273 File "/rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 143, in test_ssl > 2274 test_connect("wss://127.0.0.1:%i/" % tcp_port, EXIT_OK, noverify) > 2275 File "/rpmbuild.xpra3-rhel7.7/BUILD/xpra-3.0.5-python2/unittests/unit/server/server_sockets_test.py", line 136, in test_connect > 2276 assert r==exit_code, "expected info client to return %s but got %s" % (exit_code, client.poll()) > 2277 AssertionError: expected info client to return 0 but got None (..) > > Any clues what might be causing this? Yes, I have seen those errors before and couldn't figure out how to reproduce them reliably, so the RPMs were built by adding "--define run_tests 0" to skip running the tests during rpmbuild. You can still run the unit tests by hand and verify that they pass. More details here: http://xpra.org/trac/ticket/2432 You can run the tests with both Python 2 and 3, with just: ./unit/run2 ./unit/run3 (on CentOS 7, only Python 2 is available) Cheers, Antoine > > Thanks, > Eric. > > _______________________________________________ > shifter-users mailing list > shifter-users at lists.devloop.org.uk > https://lists.devloop.org.uk/mailman/listinfo/shifter-users >