Re: [PATCH] dpkg-base: Warn about unmounting problems

[Jan Kiszka <jan.kiszka@siemens.com>]

On 02.10.19 09:29, [ext] Jan Kiszka wrote:
> On 02.10.19 09:12, Baurzhan Ismagulov wrote:
>> From: Baurzhan Ismagulov <ibr@ilbers.de>
>>
>> If a problem occurs, log umount's error messages and warn the user about
>> retrying.
>>
>> Signed-off-by: Baurzhan Ismagulov <ibr@ilbers.de>
>> ---
>>   meta/classes/dpkg-base.bbclass | 6 ++++--
>>   1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/meta/classes/dpkg-base.bbclass b/meta/classes/dpkg-base.bbclass
>> index 2e8a0d1..42eca62 100644
>> --- a/meta/classes/dpkg-base.bbclass
>> +++ b/meta/classes/dpkg-base.bbclass
>> @@ -1,5 +1,6 @@
>>   # This software is a part of ISAR.
>>   # Copyright (C) 2017-2019 Siemens AG
>> +# Copyright (C) 2019 ilbers GmbH
>>   #
>>   # SPDX-License-Identifier: MIT
>> @@ -78,8 +79,9 @@ dpkg_do_mounts() {
>>   }
>>   dpkg_undo_mounts() {
>> -    while ! sudo umount ${BUILDROOT} 2>/dev/null; do
>> -        sleep 0.1
>> +    while ! sudo umount ${BUILDROOT}; do
>> +        bbwarn "${BUILDROOT}: Couldn't unmount, retrying..."
>> +        sleep 1
> 
> This needlessly extends the waiting time by using a course-grained delay. And 
> let's only warn warn after a some seconds.

To make it clearer: I've seen delays (i.e. failures before this patch), but I 
haven't measured them. I could imagine, though, that most of them are short 
races. And then a warning should only be issued when the delays become 
excessive, maybe 10 seconds.

Jan

-- 
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux