delta/openstack/ironic-python-agent.git/ironic_python_agent/shell/write_image.sh, branch stable/ussuri

Limit qemu-img execution arenas

2021-06-01T19:33:23+00:00

qemu-img attempts to launch multiple threads by default *and*
attempts to have multiple memory allocation arenas to operate
from. While multithreading can be good for performance, this
pattern and the memory footprint for process launch and
dependencies can turn the memory footprint for a cirros image
conversion (16MB) into 1.2GB of memory being asked for by the
qemu-img tool.

In order to limit this impact, as the default number of arenas
is governed by the number of CPUs times the number 8, it seems
reasonable to lower this to a more reasonable number which
also helps keep our possible memory footprint from being exceeded.

NOTE: This change is largely different than the original change
as an intermediate change converted write_image.sh to python.
As it is unlikely for us to backport the intermediate change,
it is logical for us to just modify the original script.
Otherwise the release note is ultimately what is backported for
release note tooling continutity.

Change-Id: I71a28ec59ec31c691205eb34d9fcab63a2ccb682
Story: 2008928
Task: 42528
(cherry picked from commit 9e4c7052a2fd9aac03858db696bf1ea9487f15e6)
(cherry picked from commit 9c20cca36284a2a17aa535bef92c828096c7d926)
(cherry picked from commit ee8118423b88f61ed58b0420356ceaed2fa53a40)

Increase the memory limit for qemu-img

2021-06-01T19:30:45+00:00

We appear to be bumping up against this limit when deploying
RHCOS images(currently 977MB). Curiously the problem isn't
happening all the time but increasing the limit eliminates it.

This limit was intruduced to guard against a malicious image
allocating an arbitrary amount of memory. Nothing else runs
on hosts when IPA is running so we should be ok bumping up
the limit.

Story: #2008667
Task: #41955
Change-Id: I9405995915a874b00b7177c9642c5469d05d66a8
(cherry picked from commit 5492ad7da59571581e9f70fd054efafb39e12593)
(cherry picked from commit de1e75126a09db9d27bfb4d77a3d2eab99c66c0c)

Clear GPT and MBR headers with dd to avoid sgdisk CRC errors

2018-08-08T16:40:22+00:00

This change adds a dd before the existing sgdisk -Z command to
workaround CRC verification errors.

Change-Id: Ia1ac4e1c0faf14ad4bb11c2a1c796c93ca8cb5e3
Closes-Bug: #1737556
Story: 1737556
Task: 11496

Fix for fatal error when GPT was used and only MBR was cleaned

2018-05-17T10:51:55+00:00

You can generate this error if after having provisioned a node
using GPT partitioning, you clean its MBR using say
dd if=/dev/zero bs=1024 count=1 of=/dev/sda
and then cleanup all Ironic/Bifrost informations to get it
reprovisioned.
In this case sgdisk -Z returns an error and last_error field
in Ironic contains:
Error writing image to device: Writing image to device
        /dev/sda failed with exit code 2
Caution: invalid main GPT header, but valid backup;
        regenerating main header\nfrom backup!\n
        \nInvalid partition data!\

Change-Id: Ib617737fff5e40cb376edda0232e0726d9c71231

Limit memory usage when running qemu-img convert

2017-01-19T18:01:30+00:00

This patch is limiting the memory address space to 1 GiB when invoking
qemu-img convert. The qemu-img command can be exploited so we better
play on the safe side (See: bug #1449062).

Change-Id: I2088f74b32d1392310c569cfaa7fd368cb737421
Closes-Bug: #1657808

Add bash scripts style checking for IPA

2016-11-09T22:14:40+00:00

Updated file tox.ini allows to check code style in all ".sh" files
included in Ironic Python Agent. Checking can be invoked by calling either
"tox" or "tox -e pep8".

Change-Id: Ie76605737c7db10a064f2aebfda388372a4c0591
Closes-Bug: #1625215

Clear GPT and MBR data structures on disk before imaging

2016-02-09T17:13:07+00:00

This patch clears all GPT and MBR data structures on the disk.
GPT maintains its data structures at start and at end of the
disk.

Change-Id: I8d0b367755b568cb44f02f62b873659b4bea0f14
Closes-Bug: #1543339

Use host_device mode for qemu-img

2015-06-18T20:00:25+00:00

When writing to a host device, possibly one that does not support holes,
it is preferable to use host_device instead of the raw mode. Not doing so
means holes are simply skipped and the deployed image is invalid.

Change-Id: I24f8b4daf099af8bb03c3a4f6417b6c87bd49021
Closes-Bug: #1466632

Write image directly to disk

2014-10-20T14:27:02+00:00

Write image directly to disk by telling qemu-img that the output
file is the disk. Still uses O_DIRECT directly within qemu-img.

This avoids the need to have a huge amount of system memory to be
able to deploy an image that contains a huge disk.

The only feature left behind by that commit is the 64K block size.

Change-Id: If60bcd82581b2ea76c4acbc3eaa3f317d2ee6590

Fix comments to reflect reality

2014-06-20T22:39:12+00:00

These comments were left in write_image.sh from the 'teeth-agent' days
when these were split into two scripts. I moved the comment into the
correct script and clarified what it means.

Change-Id: Iccdcd925be5c94d993d8125534264f496de4594f