Bug 190680
Summary: | 256MiB RAM is not enough for installing xen guest via VNC | ||
---|---|---|---|
Product: | [Fedora] Fedora Documentation | Reporter: | Robin Green <greenrd> |
Component: | release-notes | Assignee: | Release Notes Tracker <relnotes> |
Status: | CLOSED RAWHIDE | QA Contact: | Karsten Wade <kwade> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | devel | CC: | mghanawi |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | x86_64 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2007-02-10 17:20:50 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Robin Green
2006-05-04 14:01:58 UTC
What are you running on the host? Is this fully reproducible (I haven't seen it...) (In reply to comment #1) > What are you running on the host? Just gnome-terminal, mplayer and the usual services. > Is this fully reproducible (I haven't seen it...) Yes. I tried running gdbserver on the guest and it failed due to bug 190810. So I tried directly attaching to the larger anaconda process (there were 2) with gdb on the guest and upped the RAM allocation from 256MB to 300MB (to make some space for gdb), and got this: (gdb) cont Continuing. Detaching after fork from child process 266. Detaching after fork from child process 267. Detaching after fork from child process 268. Detaching after fork from child process 269. Detaching after fork from child process 270. Detaching after fork from child process 271. Detaching after fork from child process 304. Detaching after fork from child process 305. Detaching after fork from child process 306. Detaching after fork from child process 307. Detaching after fork from child process 308. Detaching after fork from child process 309. Program terminated with signal SIGKILL, Killed. The program no longer exists. (gdb) install exited abnormally Terminated sh-3.1# sending termination signals...done sending kill signals...done disabling swap... unmounting filesystems... /mnt/runtime done disabling /dev/loop0 /proc done /dev/pts done /sys done /tmp/ramfs done /selinux done you may safely reboot your system Maybe it's getting killed by the OOM killer? Maybe 256MB is not enough for an x86_64 install? (In reply to comment #2) > Maybe it's getting killed by the OOM killer? Maybe 256MB is not enough for an > x86_64 install? I was right. 256MiB, which is what I tried before, is not enough memory for an x86_64 install (at least for a VNC install on Xen). I added another memory module and retried, giving 512MiB to the xenU domain this time, and this time the install completed successfully. Reassigning to Fedora Documentation, because this is a bug in the release notes, which say: "Minimum RAM for graphical: 256MiB" There is a new section in the FC5 errata release notes that says: "At least 256 MiB of RAM for each guest, plus at least 256 MiB ram for the host" The directions you are referring to are for a standard installation, and the minimum rules must not be applied to Xen. Because I have to push up some other changes anyway, I'm adding a note in the Virtualization section to increase memory for the guest if you receive out of memory errors or as a troubleshooting step. I think it's impossible for an install to work with only 256MiB. The filesystem creation process uses tons of memory, for some reason. I wasn't clear from what you said above; are you saying you had a total of 256 MiB in the entire system, or 256 MiB designated for the Xen guest? If the system total was 256, then I agree that it wasn't enough, and fortunately, so does the release notes: http://fedora.redhat.com/docs/release-notes/fc5/#id2885418 That is the line I updated, which should be live in about 10 minutes, to say that you may need more RAM if you get OOM errors or otherwise are troubleshooting a problematic installation. If you are saying that you did follow those guidelines and had 256 each for the host and guest ... then we need input from a Xen hacker about these values, if they should be bumped, etc. Thanks. I had the same issue when assigning 256Mb to a guest during installation (through virt-manager), until I used 512Mb the installation went through without crashing. This seems to happen all the time. I also had problems installing a fully virtualized guest, I cannot seem to get passed the initial setup, the system crashes when it detects the local hard drive. The release notes now link out to a release-specific guide to using Xen: http://fedoraproject.org/wiki/FedoraXenQuickstartFC6 When so many details are needed, it becomes time to move the content from the release notes and into a stand-alone document. Thanks for the report. :) |