[yocto-autobuilder2][PATCH] builders.py: Add canStartBuild disk space check

[yocto-autobuilder2][PATCH] builders.py: Add canStartBuild disk space check

[Trevor Gamblin]

We need a way to limit the builds for when a given worker has less than
a certain amount of disk space available. This implements a
canStartBuild method based on the example in the Buildbot docs and
blocks a build if the worker has less than 60GB of disk space available.
Unlike the example code, we want the stdout of the command so that we
can calculate the amount of disk space, rather than just relying on the
remote command's return code.

Docs: https://docs.buildbot.net/latest/manual/customization.html#canstartbuild-functions

[YOCTO #14591]

Signed-off-by: Trevor Gamblin <trevor.gamblin@windriver.com>
---
 builders.py | 44 ++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 40 insertions(+), 4 deletions(-)

diff --git a/builders.py b/builders.py
index 5773950..0d1facc 100644
--- a/builders.py
+++ b/builders.py
@@ -8,6 +8,7 @@ from yoctoabb import config
 from yoctoabb.steps.writelayerinfo import WriteLayerInfo
 from yoctoabb.steps.runconfig import get_publish_dest, get_publish_resultdir, get_publish_name, RunConfigCheckSteps, TargetPresent
 from buildbot.process.results import Results, SUCCESS, FAILURE, CANCELLED, WARNINGS, SKIPPED, EXCEPTION, RETRY
+from buildbot.process.remotecommand import RemoteCommand
 
 from twisted.python import log
 from twisted.internet import defer
@@ -45,6 +46,41 @@ def ensure_props_set(props):
         "publish_destination": props.getProperty("publish_destination", "")
     }
 
+@defer.inlineCallbacks
+def shell(command, worker, builder):
+    args = {
+        'command': command,
+        'workdir': worker.worker_basedir,
+        'logEnviron': False,
+        'want_stdout': True,
+        'want_stderr': False,
+    }
+
+    cmd = RemoteCommand('shell', args, collectStdout=True, stdioLogName="stdio")
+    cmd.worker = worker
+    yield cmd.run(None, worker.conn, builder.name)
+    return cmd
+
+@defer.inlineCallbacks
+def canStartBuild(builder, wfb, request):
+    log.msg("Checking available disk space...")
+
+    cmd = yield shell("df -BG | grep $(findmnt -T . | awk '{print $2}' | sed -n 2p) | awk '{print $4}' | sed 's/[^0-9]*//g'", wfb.worker, builder)
+    threshold = 60 # GB of space
+    if int(cmd.stdout) < threshold:
+        log.msg("Detected {0} GB of space available, less than threshold of {1} GB. Can't start build".format(cmd.stdout, threshold))
+        wfb.worker.putInQuarantine()
+        return False
+    else:
+        log.msg("Detected {0} GB of space available, more than threshold of {1} GB. OK to build".format(cmd.stdout, threshold))
+
+    wfb.worker.quarantine_timeout = 120
+    wfb.worker.putInQuarantine()
+
+    wfb.worker.resetQuarantine()
+
+    return True
+
 def create_builder_factory():
     f = util.BuildFactory()
 
@@ -136,7 +172,7 @@ for builder in config.subbuilders:
     if not workers:
         workers = config.builder_to_workers['default']
     builders.append(util.BuilderConfig(name=builder,
-                                       workernames=workers, nextWorker=nextWorker, nextBuild=nextBuild,
+                                       workernames=workers, canStartBuild=canStartBuild, nextWorker=nextWorker, nextBuild=nextBuild,
                                        factory=f, env=extra_env))
 
 # Prioritize assigning builders to available workers based on the length
@@ -300,8 +336,8 @@ def create_parent_builder_factory(buildername, waitname):
 
     return factory
 
-builders.append(util.BuilderConfig(name="a-quick", workernames=config.workers, factory=create_parent_builder_factory("a-quick", "wait-quick"), nextWorker=nextWorker, nextBuild=nextBuild, env=extra_env))
-builders.append(util.BuilderConfig(name="a-full", workernames=config.workers, factory=create_parent_builder_factory("a-full", "wait-full"), nextWorker=nextWorker, nextBuild=nextBuild, env=extra_env))
+builders.append(util.BuilderConfig(name="a-quick", workernames=config.workers, factory=create_parent_builder_factory("a-quick", "wait-quick"), canStartBuild=canStartBuild, nextWorker=nextWorker, nextBuild=nextBuild, env=extra_env))
+builders.append(util.BuilderConfig(name="a-full", workernames=config.workers, factory=create_parent_builder_factory("a-full", "wait-full"), canStartBuild=canStartBuild,nextWorker=nextWorker, nextBuild=nextBuild, env=extra_env))
 
 def create_doc_builder_factory():
     f = util.BuildFactory()
@@ -345,4 +381,4 @@ def create_doc_builder_factory():
 
 # Only run one docs build at a time
 docs_lock = util.MasterLock("docs_lock")
-builders.append(util.BuilderConfig(name="docs", workernames=config.workers, factory=create_doc_builder_factory(), nextWorker=nextWorker, nextBuild=nextBuild, env=extra_env, locks=[docs_lock.access('exclusive')]))
+builders.append(util.BuilderConfig(name="docs", workernames=config.workers, factory=create_doc_builder_factory(), canStartBuild=canStartBuild, nextWorker=nextWorker, nextBuild=nextBuild, env=extra_env, locks=[docs_lock.access('exclusive')]))
-- 
2.31.1

Re: [yocto] [yocto-autobuilder2][PATCH] builders.py: Add canStartBuild disk space check

[Richard Purdie]

On Fri, 2021-10-29 at 09:22 -0400, Trevor Gamblin wrote:
> We need a way to limit the builds for when a given worker has less than
> a certain amount of disk space available. This implements a
> canStartBuild method based on the example in the Buildbot docs and
> blocks a build if the worker has less than 60GB of disk space available.
> Unlike the example code, we want the stdout of the command so that we
> can calculate the amount of disk space, rather than just relying on the
> remote command's return code.
> 
> Docs: https://docs.buildbot.net/latest/manual/customization.html#canstartbuild-functions
> 
> [YOCTO #14591]
> 
> Signed-off-by: Trevor Gamblin <trevor.gamblin@windriver.com>
> ---
>  builders.py | 44 ++++++++++++++++++++++++++++++++++++++++----
>  1 file changed, 40 insertions(+), 4 deletions(-)
> 
> diff --git a/builders.py b/builders.py
> index 5773950..0d1facc 100644
> --- a/builders.py
> +++ b/builders.py
> @@ -8,6 +8,7 @@ from yoctoabb import config
>  from yoctoabb.steps.writelayerinfo import WriteLayerInfo
>  from yoctoabb.steps.runconfig import get_publish_dest, get_publish_resultdir, get_publish_name, RunConfigCheckSteps, TargetPresent
>  from buildbot.process.results import Results, SUCCESS, FAILURE, CANCELLED, WARNINGS, SKIPPED, EXCEPTION, RETRY
> +from buildbot.process.remotecommand import RemoteCommand
>  
>  from twisted.python import log
>  from twisted.internet import defer
> @@ -45,6 +46,41 @@ def ensure_props_set(props):
>          "publish_destination": props.getProperty("publish_destination", "")
>      }
>  
> +@defer.inlineCallbacks
> +def shell(command, worker, builder):
> +    args = {
> +        'command': command,
> +        'workdir': worker.worker_basedir,
> +        'logEnviron': False,
> +        'want_stdout': True,
> +        'want_stderr': False,
> +    }
> +
> +    cmd = RemoteCommand('shell', args, collectStdout=True, stdioLogName="stdio")
> +    cmd.worker = worker
> +    yield cmd.run(None, worker.conn, builder.name)
> +    return cmd
> +
> +@defer.inlineCallbacks
> +def canStartBuild(builder, wfb, request):
> +    log.msg("Checking available disk space...")
> +
> +    cmd = yield shell("df -BG | grep $(findmnt -T . | awk '{print $2}' | sed -n 2p) | awk '{print $4}' | sed 's/[^0-9]*//g'", wfb.worker, builder)
> +    threshold = 60 # GB of space
> +    if int(cmd.stdout) < threshold:
> +        log.msg("Detected {0} GB of space available, less than threshold of {1} GB. Can't start build".format(cmd.stdout, threshold))
> +        wfb.worker.putInQuarantine()
> +        return False
> +    else:
> +        log.msg("Detected {0} GB of space available, more than threshold of {1} GB. OK to build".format(cmd.stdout, threshold))
> +
> +    wfb.worker.quarantine_timeout = 120
> +    wfb.worker.putInQuarantine()
> +
> +    wfb.worker.resetQuarantine()
> +
> +    return True
> +

Unfortunately this quarantine piece is causing problems. It means that even if
there is free space on the builder and always was, a maximum of one build every
2 minutes can be started.

I'll probably drop the quarantine piece as it was a nice to have soft start for
recovery rather than an essential.

Cheers,

Richard