Hey there,

We need to see the pstacks from all threads to really determine the cause here. 
Can you send us a complete read out? 

gdb -p "pid"

thread apply all bt


It'd be great if you can install debug info too to help. 

Thanks,


> On 31 May 2020, at 05:09, Crocker, Deborah <[email protected]> wrote:
> 
> Some more information from a coworker:
> 
> Yeah, this sounds like an LDAP server bug. I haven't figured out what to look 
> at to pin it down, but when it's slow to connect, I can see with strace that 
> the primary thread hasn't called accept() yet for the connection I'm trying 
> to open. Once it does, the whole thing goes very quickly, and I usually see a 
> burst of other connections accepted and handled at the same time.
> 
> Deborah Crocker, PhD
> Systems Engineer III 
> Office of Information Technology 
> The University of Alabama
> Box 870346 
> Tuscaloosa, AL 36587 
> Office 205-348-3758 | Fax 205-348-9393 
> [email protected]
> 
> 
> -----Original Message-----
> From: Crocker, Deborah <[email protected]> 
> Sent: Saturday, May 30, 2020 2:08 PM
> To: General discussion list for the 389 Directory server project. 
> <[email protected]>
> Subject: [EXTERNAL] [389-users] new server setup hanging
> 
> We're trying to move into our new server setup. We have one that seems to be 
> fine under a load but when we bring the next we're having trouble with it 
> hanging. The second does have more clients (and different) so there could be 
> something about what a client is doing. Here is the server:
>       389-Directory/1.3.10.1 B2020.133.1625
> Installed from EPEL, running on
>       CentOS Linux release 7.8.2003
> 
> And here is the pstack output listing the only thread that is not idle. Can 
> anyone tell me what is going on?
> 
> Thread 44 (Thread 0x7f858e9b3700 (LWP 2515)):
> #0  0x00007f860a90fe02 in slapi_atomic_load_32 () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #1  0x00007f860a8d4e8e in slapi_get_mapping_tree_node_by_dn () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #2  0x00007f860a8d5179 in slapi_be_select () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #3  0x00007f860a9296a0 in vattr_test_filter () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #4  0x00007f860a8b6ec4 in slapi_vattr_filter_test_ext_internal () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #5  0x00007f860a8b7ba6 in slapi_vattr_filter_test_ext () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #6  0x00007f8600a99e02 in acl__resource_match_aci () at 
> /usr/lib64/dirsrv/plugins/libacl-plugin.so
> #7  0x00007f8600a9b280 in acl_access_allowed () at 
> /usr/lib64/dirsrv/plugins/libacl-plugin.so
> #8  0x00007f8600aae9f7 in acl_access_allowed_main () at 
> /usr/lib64/dirsrv/plugins/libacl-plugin.so
> #9  0x00007f860a8f0cbc in plugin_call_acl_plugin () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #10 0x00007f860a8b638d in test_filter_access () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #11 0x00007f860a8b6fb5 in slapi_vattr_filter_test_ext_internal () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #12 0x00007f860a8b6d31 in slapi_vattr_filter_test_ext_internal () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #13 0x00007f860a8b7ba6 in slapi_vattr_filter_test_ext () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #14 0x00007f85ff7c0df1 in ldbm_back_next_search_entry_ext () at 
> /usr/lib64/dirsrv/plugins/libback-ldbm.so
> #15 0x00007f860a8deca6 in send_results_ext.constprop.5 () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #16 0x00007f860a8e0e09 in op_shared_search () at 
> /usr/lib64/dirsrv/libslapd.so.0
> #17 0x0000557410dd3c0e in do_search ()
> #18 0x0000557410dc198a in connection_threadmain ()
> #19 0x00007f86086a0c5b in _pt_root () at /lib64/libnspr4.so
> #20 0x00007f8608040ea5 in start_thread () at /lib64/libpthread.so.0
> #21 0x00007f86076ec8dd in clone () at /lib64/libc.so.6
> 
> Deborah Crocker, PhD
> Systems Engineer III 
> Office of Information Technology 
> The University of Alabama
> Box 870346 
> Tuscaloosa, AL 36587 
> Office 205-348-3758 | Fax 205-348-9393 
> [email protected]
> 
> 
> -----Original Message-----
> From: William Brown <[email protected]> 
> Sent: Wednesday, May 27, 2020 5:43 PM
> To: [email protected]
> Subject: [EXTERNAL] [389-users] Re: Re: Re: Advice to bring new servers into 
> production
> 
> 
> 
>> On 27 May 2020, at 23:20, Crocker, Deborah <[email protected]> wrote:
>> 
>> Thanks - I think we have enough ideas in here to get this going. One last 
>> question: 
>> If replication is set up through the host name - how often does the 
>> directory server do a DNS look up, or does it do it once on startup (or 
>> creation of the rep agreement)?
> 
> I "think" it's every time it initiates the new connection - but remember, for 
> replication, that *is* quite different to a client doing a search, so I'd be 
> pretty careful about this. IMO you should be standing up your replacement 
> servers in parallel, joining them all, moving the IP's then decomission the 
> old servers. Alternately, you'll need an outage window to shutdown your old 
> servers, export the ldif, and then import and bring up the new ones.
> 
> I think having "IP's are a limited resource" really does make this whole 
> process much much harder than it needs to be for you ... :( 
> 
>> 
>> -----Original Message-----
>> From: William Brown <[email protected]>
>> Sent: Tuesday, May 26, 2020 10:48 PM
>> To: [email protected]
>> Subject: [EXTERNAL] [389-users] Re: Re: Advice to bring new servers 
>> into production
>> 
>> There are a few options. The best would be a load balancer which has the 
>> ip's so that it's transparent to your LDAP servers where they are.
>> 
>> But also as mentioned, the virtual IP's honestly is the best way. Linux can 
>> have multiple IP's on an interface so you can just have two IP's on one 
>> interface, andthat's the best way to do this.
>> 
>> Alternately, don't rely on the IP, lower your DNS ttl's to a very short 
>> time, change the DNS A/AAAA records, and then do it that way.
>> 
>> 
>> 
>>> On 27 May 2020, at 06:17, Crocker, Deborah <[email protected]> wrote:
>>> 
>>> I’d like not to take up two ip addresses per host indefinitely. We have 
>>> re-IP’d our hosts before so I know we can to do this but it was during a 
>>> downtime when everything was restarted. Just trying to get away with not 
>>> restarting the masters.
>>> 
>>> Deborah Crocker, PhD
>>> Systems Engineer III
>>> Office of Information Technology
>>> The University of Alabama
>>> Box 870346
>>> Tuscaloosa, AL 36587
>>> Office 205-348-3758 | Fax 205-348-9393 [email protected]
>>> 
>>> From: Leo Pleiman <[email protected]>
>>> Sent: Tuesday, May 26, 2020 3:08 PM
>>> To: General discussion list for the 389 Directory server project. 
>>> <[email protected]>
>>> Subject: [EXTERNAL] [389-users] Re: Advice to bring new servers into 
>>> production
>>> 
>>> My experience has been that the replicas and consumers have a unique id, 
>>> more than just an IP address which creates the trust relationship with the 
>>> master. If your goal is to simply maintain an IP so your clients don't have 
>>> to be repointed, I would build each new LDAP host and replication 
>>> agreement, and then as you decommission the old hosts use their IP address 
>>> as a virtual IP address on the replacement host. It would take a quick 
>>> restart od the LDAP service to start a listener on the virtual Ip address.
>>> 
>>> 
>>> Leo Pleiman
>>> Senior System Engineer
>>> Direct 202-787-3622
>>> Cell 410-688-3873
>>> 
>>> 
>>> 
>>> On Tue, May 26, 2020 at 3:57 PM Crocker, Deborah <[email protected]> wrote:
>>> We have a setup with 2 multi-masters and 3 consumers. We are now building 
>>> new host and want to put them in place ultimately at the same IP address as 
>>> the original ones. I need some advice on how to do this quickly and cleanly.
>>> 
>>> To add a new consumer the idea now is to set it up and set up replications 
>>> agreements from each master using consumer DNS name (don't start continuous 
>>> replication yet). After initializing new consumer from one master - turn 
>>> off old consumer, remove old consumer agreement from each master, and re-IP 
>>> new consumer. Do we need to restart masters to re-read DNS or will it pick 
>>> that up when it starts the next replication? Is this the best way to do 
>>> this?
>>> 
>>> Thanks
>>> 
>>> Deborah Crocker, PhD
>>> Systems Engineer III
>>> Office of Information Technology
>>> The University of Alabama
>>> Box 870346
>>> Tuscaloosa, AL 36587
>>> Office 205-348-3758 | Fax 205-348-9393 [email protected]
>>> 
>>> _______________________________________________
>>> 389-users mailing list -- [email protected] To 
>>> unsubscribe send an email to [email protected]
>>> Fedora Code of Conduct: 
>>> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
>>> List Guidelines: 
>>> https://fedoraproject.org/wiki/Mailing_list_guidelines
>>> List Archives: 
>>> https://lists.fedoraproject.org/archives/list/[email protected]
>>> r oject.org _______________________________________________
>>> 389-users mailing list -- [email protected] To 
>>> unsubscribe send an email to [email protected]
>>> Fedora Code of Conduct: 
>>> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
>>> List Guidelines: 
>>> https://fedoraproject.org/wiki/Mailing_list_guidelines
>>> List Archives: 
>>> https://lists.fedoraproject.org/archives/list/[email protected]
>>> r
>>> oject.org
>> 
>> —
>> Sincerely,
>> 
>> William Brown
>> 
>> Senior Software Engineer, 389 Directory Server SUSE Labs 
>> _______________________________________________
>> 389-users mailing list -- [email protected] To 
>> unsubscribe send an email to [email protected]
>> Fedora Code of Conduct: 
>> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
>> List Guidelines: 
>> https://fedoraproject.org/wiki/Mailing_list_guidelines
>> List Archives: 
>> https://lists.fedoraproject.org/archives/list/[email protected]
>> oject.org _______________________________________________
>> 389-users mailing list -- [email protected] To 
>> unsubscribe send an email to [email protected]
>> Fedora Code of Conduct: 
>> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
>> List Guidelines: 
>> https://fedoraproject.org/wiki/Mailing_list_guidelines
>> List Archives: 
>> https://lists.fedoraproject.org/archives/list/[email protected]
>> oject.org
> 
> —
> Sincerely,
> 
> William Brown
> 
> Senior Software Engineer, 389 Directory Server SUSE Labs 
> _______________________________________________
> 389-users mailing list -- [email protected] To unsubscribe 
> send an email to [email protected]
> Fedora Code of Conduct: 
> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: 
> https://lists.fedoraproject.org/archives/list/[email protected]
> _______________________________________________
> 389-users mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> Fedora Code of Conduct: 
> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: 
> https://lists.fedoraproject.org/archives/list/[email protected]
> _______________________________________________
> 389-users mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> Fedora Code of Conduct: 
> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: 
> https://lists.fedoraproject.org/archives/list/[email protected]

—
Sincerely,

William Brown

Senior Software Engineer, 389 Directory Server
SUSE Labs
_______________________________________________
389-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/[email protected]

Reply via email to