Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    _ChezW@m_
    @JrmChP_twitter
    ok thanks
    _ChezW@m_
    @JrmChP_twitter
    other inventories question, is the /var/rudder/inventories/received/ folder supposed to be cleared ? mine have about 1100 files, with the older from 2018? Half of then are from today.
    Francois Armand
    @fanf
    @JrmChP_twitter no, received is for all received inventories.
    Francois Armand
    @fanf
    That case is still disturbing. @JrmChP_twitter : would you be ok to share with me your 1000 inventories/.sign files? I would like to try to reproduce the behavior and find the root cause (and a real fix)
    _ChezW@m_
    @JrmChP_twitter
    hello @fanf yes, no pb.
    i've moved to another hypervisor this morning, i'll see if it's better.
    Francois Armand
    @fanf
    @JrmChP_twitter thanks, perfect, I will try to reproduce
    Francois Armand
    @fanf

    @JrmChP_twitter so, on a test plateform, I have not good restsults with 1000 nodes, but still order better than yours: 

    [2021-01-12 13:42:10+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 18948 ms
[2021-01-12 13:42:27+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 8585 ms
[2021-01-12 13:42:34+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 7202 ms
[2021-01-12 13:42:46+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 7463 ms
[2021-01-12 13:42:53+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 2942 ms
[2021-01-12 13:43:11+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 18257 ms
[2021-01-12 13:43:22+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 2854 ms
[2021-01-12 13:43:26+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 3224 ms

    I didn't encounter any dead lock though, which is good (even if it may be just because I wasn't able to get to the broken state).
    I will try to make the merge_uuid part better and see if it solves your problem.

    _ChezW@m_
    @JrmChP_twitter
    ok thanks
    Suvi
    @Suvi8

    Hello guys

    How can i make a check if the file does not exist? I see only File check exists.

    Francois Armand
    @fanf
    @JrmChP_twitter so, first thing to do: in /opt/rudder/etc/openldap/slapd.conf: comment index software and modifyTimestamp, then: 
    systemctl stop rudder-slapd
su - rudder-slapd -s /bin/sh -c "/opt/rudder/sbin/slapindex -v -t -q" 
systemctl start rudder-slapd
    _ChezW@m_
    @JrmChP_twitter
    Hello @fanf , it's done
    should i re enable index ?
    Suvi
    @Suvi8

    Hello guys, sorry to ask you again.

    How can i make a check if the file does not exist? I see only File check exists technique.

    Francois Armand
    @fanf
    @JrmChP_twitter no, they will need to remain disable. Does it make things better?
    Fdall
    @Fdall
    @Suvi8 Hi, I am not sure of what you want to do. Do you want to have a success report is the file does not exists, and an error one when it does? or do you want to check its existence to define conditions to guard others configuration points?
    _ChezW@m_
    @JrmChP_twitter
    ok @fanf i watch the inventory perf and get back to you
    Suvi
    @Suvi8
    yes exactly i want to make a condition, for example: if file doesnt exist, then apply condition xyz
    _ChezW@m_
    @JrmChP_twitter

    yes exactly i want to make a condition, for example: if file doesnt exist, then apply condition xyz

    you can maybe test it with a single command ? like a simple ls ?

    Fdall
    @Fdall
    Ok, your best bet it to use the method called "File check exists" in the technique where you will use that condition. To use the resulting condition just copy/paste the success line in the "result condition" tab of the File check exists method after you fulfilled its parameters
    The method will check by herself if the file exists, there is no need to try using cli execution here
    Suvi
    @Suvi8
    but if file check doesnt exist, it will show me as error in compliance...
    Vincent Membré
    @VinceMacBuche
    but on second run, when file will be created (if you do so with the negative condition) you will have a success
    Fdall
    @Fdall
    @VinceMacBuche File will not be created, the method just apply a check and I believe this is a case where the configuration to deploty depends on the current state on the machine and must not be "repaired" by the agent
    Vincent Membré
    @VinceMacBuche
    depends if the file is created in the next steps of the technique
    if depends of a state of the machine you should do it via properties or using "node state" and using groups to apply it or not
    Fdall
    @Fdall
    @Suvi8 If you really don't want to have an error when the file do not exists, currently you have to use the "condition_from_command" which should always return a success and define some special conditions described in its documentation regarding the return code of the command exit code.
    _ChezW@m_
    @JrmChP_twitter

    @fanf still have the same numbers of inventories in folders: 

    root@rudder:/var/rudder/inventories# l accepted-nodes-updates/ |wc -l
66
root@rudder:/var/rudder/inventories# l incoming/ |wc -l
17
root@rudder:/var/rudder/inventories# l received/ |wc -l
1095

    are you interested in other values ?

    Francois Armand
    @fanf
    @JrmChP_twitter does the timing of Precommit 'pre_commit_inventory:merge_uuid': are better ?
    _ChezW@m_
    @JrmChP_twitter
    @fanf i've put trace back: 
    root@rudder:~# tailf /opt/rudder/etc/logback.xml  /var/log/rudder/webapp/2021_01_13.stderrout.log |grep "pre_commit_inventory:merge_uuid"
[2021-01-13 13:10:10+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 41332 ms
[2021-01-13 13:13:08+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 171481 ms
    and before: 
    [2021-01-11 01:03:11+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 253823 ms
[2021-01-11 01:08:26+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 299530 ms
[2021-01-11 01:14:24+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 338230 ms
[2021-01-11 01:19:52+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 312329 ms
[2021-01-11 01:25:48+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 343000 ms
[2021-01-11 01:26:18+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 9700 ms
    i let it few minutes more, i let you know
    _ChezW@m_
    @JrmChP_twitter
    [2021-01-13 13:24:52+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 234982 ms
[2021-01-13 13:29:03+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 241389 ms
[2021-01-13 13:32:07+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 173675 ms
[2021-01-13 13:36:00+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 221762 ms
[2021-01-13 13:39:32+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 201154 ms
[2021-01-13 13:43:48+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 245041 ms
[2021-01-13 13:47:08+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 188606 ms
[2021-01-13 13:50:36+0100] TRACE inventory-logger.timing - Precommit 'pre_commit_inventory:merge_uuid': 190892 ms
    Francois Armand
    @fanf
    @JrmChP_twitter I believe I identified the root cause. There's a pb with the merge of software causing an explosion of their number (and indexes are a second problem, but not yet sure how they interact). I'm working on a patch
    _ChezW@m_
    @JrmChP_twitter
    ok good luck with that! Thanks @fanf
    Francois Armand
    @fanf
    @JrmChP_twitter OK, done. It will converge after having inventories for all nodes. And I added two things:
    • an API endpint to trigger software cleaning ( POST on .../api/latest/system/maintenance/purgeSoftware)
    • priorisation of inventories for new nodes compare to update to update to already accepted (so that you don't wait a long time before a new node appears)
    Francois Armand
    @fanf
    https://github.com/Normation/rudder/pull/3472/files
    _ChezW@m_
    @JrmChP_twitter
    ok. How should i push it to my server? wait for new release?
    Francois Armand
    @fanf
    @JrmChP_twitter let me see today. It will be in next nightly build, but we are planning to do a release ASAP to correct that very serious problem
    _ChezW@m_
    @JrmChP_twitter
    cool
    :)
    _ChezW@m_
    @JrmChP_twitter
    if i need an host to be registered quickly, is there a way to force the inventory to be treated ?
    _ChezW@m_
    @JrmChP_twitter
    forget it, ok by moving incoming/ and accepted-nodes-updates/ on a temp folder, and let just my node files.
    Francois Armand
    @fanf
    :thumbsup:
    Suvi
    @Suvi8
    @Fdall Can you send me a printscreen what should i apply in my technique?
    Fdall
    @Fdall
    @Suvi8 I send you a pm since the explanations are a bit long
    Matthew Frost
    @mattronix
    whats the best way to add a PPa with rudder
    command execution?
    Alexis Mousset
    @amousset
    This, or copy files directly into /etc/apt/sources.list.d
    Matthew Frost
    @mattronix
    ah clear