Red Hat Bugzilla – Bug 596981
one minute delay from "Found installation media" to "running anaconda"
Last modified: 2010-07-31 11:42:24 EDT
Description of problem: At the beginning of a fresh install from DVD, it takes almost one minute to get from "Found installation media" to "running anaconda 14.6", and another 20 seconds before X11 starts. The total delay of around 80 seconds is too slow by a factor of ten. During this time the VGA screen shows a plain blue background with some navigation text on the top and bottom lines.
The activity light on the DVD drive is on (indicating active operation) 99% of the time. There are no sounds of frequent seeking. It is a 4X DVD+RW and can deliver 5MB/second. What requires 300MB of data at this stage?
Version-Release number of selected component (if applicable):
How reproducible: every time
Steps to Reproduce:
1. Compose install DVD from rawhide using pungi; burn to DVD+RW.
2. Boot DVD for fresh install; Skip media check
Actual results: About 55 to 58 seconds after Skip media check, from "Found installation media" to "running anaconda 14.6". Another 20 seconds until the screen shows X11 starting.
Expected results: No more than 6 or 7 seconds from "Found installation media" to "running anaconda".
What's /tmp/syslog say? Are you getting a lot of read errors?
/tmp/syslog is unavailable because the system hangs/crashes at start of X11: bug
596985. VT2 has no shell: too early.
"dd if=/dev/sr0 bs=32k count=10000 of=/dev/null" says 330MB in 106 seconds (3.5 MB/s) with no errors in /var/log/messages, when run on an already-installed F14 rawhide on the same box. Media Check also passes when run during another boot on the same box. Drive is Optiarc AD-7240S about 8 months old, lightly used, and is the drive that burned the DVD in the first place, earlier today.
Well, you could always try starting up a text install, seeing if you get the same long delay, and then grabbing the information if so. Or you could try the same thing with a serial console install, or a VNC install. anaconda supports syslog=. There's plenty of ways to get information.
Running a text install (append " text" to the end of the kernel boot command line), then there is no file whose pathname matches /tmp/*syslog*. Perhaps you mean /var/log/messages? That is the filename which appears in "ls -l /proc/$(< /var/run/syslogd.pid)/fd". Anyway there aren't any I/O errors listed in /var/log/messages.
What is the complete syntax that is implied by "anaconda supports syslog="? What other pieces must be available (such as a path through a firewall on another machine?), and how is that setup?
> What is the complete syntax that is implied by "anaconda supports syslog="?
> What other pieces must be available (such as a path through a firewall on
> another machine?), and how is that setup?
This bug appears to have been reported against 'rawhide' during the Fedora 14 development cycle.
Changing version to '14'.
More information and reason for this action is here:
(In reply to comment #5)
Some of the info on that page does not work. In particular, there is no file 'analog' anywhere in anaconda-13.42-1, which is the current shipping release. Thus it is problematic to construct the syslog configuration.
Found a system with anaconda-14.12 that could run the 'analog' script to generate the syslogd configuration, and transfered the configuration file to the logging machine, and started the logger, and appended " syslog=<ip4>:<port>" to the kernel boot command line.
There are no suspicious messages in syslog, but the one-minute delay no longer appears, either.