Re: Error from suggester: "Searching for Solr?"

[Mike]

Thank you for taking the trouble.

I took the Solrconfig.xml from _default and took over the suggester from
techproducts example with three changes.
Unfortunately, I continue to get the error and the suggester is also not
displayed as a plugin.

https://paste.debian.net/1267534/

Does the suggester basically work with a nutch scheme?

Thanks

Mike

Am Mo., 16. Jan. 2023 um 22:14 Uhr schrieb Shawn Heisey :

> On 1/16/2023 9:25 AM, Mike wrote:
> > Thank you for your help.
> > I don't see any errors in the logs.
> >
> > https://paste.debian.net/hidden/89df3382/
> >
> > mike
>
> If I try with your solrconfig.xml, after fixing errors about missing
> fieldTypes in the schema, I see that the /suggest handler is not
> present.  The XML looks right to me and passes validation, so I can't
> figure out why it doesn't work.
>
> When I just added the two sections to my own solrconfig.xml, the handler
> was present.
>
> Thanks,
> Shawn
>
>

General question about high availability.

[Matthew Castrigno]

What is the best approach for high availability for my deployed SOLR instance? 
Solr clusters seems to be for scaling, which is not my concern as my use case 
is a simple site search.

This SOLR instance is expected to serve about 150K requests per month. The 
documents are relatively small with about 50 fields indexed each.

Thank you for your insights, they are appreciated.


[cid:38f99cd3-e372-41c1-8669-022d778a9338]

Matthew Castrigno

IHT Developer II

St. Luke’s Health System

•  208-859-4276
•  castr...@slhs.org

--
"This message is intended for the use of the person or entity to which it is 
addressed and may contain information that is confidential or privileged, the 
disclosure of which is governed by applicable law. If the reader of this 
message is not the intended recipient, you are hereby notified that any 
dissemination, distribution, or copying of this information is strictly 
prohibited. If you have received this message by error, please notify us 
immediately and destroy the related message."

Re: General question about high availability.

[Markus Jelsma]

Hello Matthew,

Solr cloud is not only for scaling (add more shards), but also for
availability (add more replicas).

As long as you place the replicas on different physical machines, or VMs on
different physical machines, or even some VMs in different DCs, you are as
high available as you need to be.

We used to place on VMs, even in different DCs, and that worked fine for
years. The only possible downtime was human error. Now we place replicas on
VMs on different physical machines in the same DC, which works just as fine.

Regards,
Markus

Op di 17 jan. 2023 om 17:06 schreef Matthew Castrigno :

> What is the best approach for high availability for my deployed SOLR
> instance? Solr clusters seems to be for scaling, which is not my concern as
> my use case is a simple site search.
>
> This SOLR instance is expected to serve about 150K requests per month. The
> documents are relatively small with about 50 fields indexed each.
>
> Thank you for your insights, they are appreciated.
>
> *Matthew Castrigno*
>
> *IHT Developer II*
>
> *St. Luke’s Health System*
>
> (  208-859-4276
> +  castr...@slhs.org 
> --
> "This message is intended for the use of the person or entity to which it
> is addressed and may contain information that is confidential or
> privileged, the disclosure of which is governed by applicable law. If the
> reader of this message is not the intended recipient, you are hereby
> notified that any dissemination, distribution, or copying of this
> information is strictly prohibited. If you have received this message by
> error, please notify us immediately and destroy the related message."
>

Re: General question about high availability.

[Dave]

Put an nginx front for about three solr servers that does a drop down failover. 
You always want one to be the primary for caching and that few searches, then 
drop down to the other couple on failure 

> On Jan 17, 2023, at 12:07 PM, Matthew Castrigno  wrote:
> 
> 
> What is the best approach for high availability for my deployed SOLR 
> instance? Solr clusters seems to be for scaling, which is not my concern as 
> my use case is a simple site search. 
> 
> This SOLR instance is expected to serve about 150K requests per month. The 
> documents are relatively small with about 50 fields indexed each.
> 
> Thank you for your insights, they are appreciated.
> 
>  
> Matthew Castrigno
> IHT Developer II
> St. Luke’s Health System
> (  208-859-4276
> +  castr...@slhs.org
> "This message is intended for the use of the person or entity to which it is 
> addressed and may contain information that is confidential or privileged, the 
> disclosure of which is governed by applicable law. If the reader of this 
> message is not the intended recipient, you are hereby notified that any 
> dissemination, distribution, or copying of this information is strictly 
> prohibited. If you have received this message by error, please notify us 
> immediately and destroy the related message."

Re: General question about high availability.

[dmitri maziuk]

On 2023-01-17 11:05 AM, Matthew Castrigno wrote:
What is the best approach for high availability for my deployed SOLR 
instance? Solr clusters seems to be for scaling, which is not my concern 
as my use case is a simple site search.


We used to have a 3-node cluster behind a L2 proxy plus a 
development/playground node. I am arguing for a single node plus a copy 
of production index on the dev (VMs cost ), probably behind an F5 
configured to forward everything to prod node unless it's down... but 
we'll see what we end up with.


Dima

Re: General question about high availability.

[Shawn Heisey]

On 1/17/23 10:05, Matthew Castrigno wrote:
What is the best approach for high availability for my deployed SOLR 
instance? Solr clusters seems to be for scaling, which is not my concern 
as my use case is a simple site search.


This SOLR instance is expected to serve about 150K requests per month. 
The documents are relatively small with about 50 fields indexed each.


If you go with SolrCloud, you need three servers minimum.  Zookeeper 
requires 3 or more for high availability.  Solr itself only requires 
two.  So the most minimal cloud setup is 3 servers all running ZK, with 
two of them also running Solr.


All servers must be on separate physical hardware.  If you create three 
VMs all on the same host, then if that host dies, your Solr is 
completely down.


Nginx was mentioned as a load balancer.  I use haproxy, but nginx would 
work well too.  It is a good idea when setting up load balancing to have 
a virtual IP address.  Two programs for setting that up are ucarp and 
pacemaker.  I'm sure there are more options.


Thanks,
Shawn

Re: General question about high availability.

[dmitri maziuk]

On 2023-01-17 1:15 PM, Shawn Heisey wrote:
...  It is a good idea when setting up load balancing to have
a virtual IP address.  Two programs for setting that up are ucarp and 
pacemaker.  I'm sure there are more options.


Not really. There used to be heartbeat but it was deprecated, gone, and 
buried by linux-ha folks because everybody needs N+M setups of I+J 
services running on A+B servers in X+Y geographically distributed 
datacenters -- pacemaker to the rescue. IMO pacemaker (well, linux-ha 
stack) is way too bloated for the simple common use cases, and ucarp... 
Well, you'd be better off running BSD where CARP is part of the kernel 
proper, if it weren't for the Java-on-BSD problem.


Better yet, run "in the cloud" and let kube worry about VIPs and 
availability and the rest of it.


Dima

Re: Logging correlation id

[Jan Høydahl]

Distributed Tracing :: Apache Solr Reference Guidesolr.apache.orgJan Høydahl16. jan. 2023 kl. 09:28 skrev DAVID MARTIN NIETO :Where is that?enabling tracing, (header trace-parent)Thanks.De: Jan Høydahl Enviado: viernes, 13 de enero de 2023 16:42Para: users@solr.apache.org Asunto: Re: Logging correlation idHave you considered enabling tracing, (header trace-parent) and get that Id added to the logs?Jan Høydahl13. jan. 2023 kl. 02:42 skrev Shawn Heisey :On 1/12/23 12:19, Srijan wrote:I am looking to log additional info to my Solr log. Specifically, I amlooking to log a "correlation-id" provided by my client as part of the HTTPheader field "x-correlation-id" and put that in my log to relate variousclient sessions and their actions with Solr. Is there a codeless way ofdoing this? Has anyone done something similar?If you make it a URL parameter instead of a header, Solr will log it automatically.Arbitrary URL parameters that do not match anything Solr expects tend to be ignored, so it is a good way to log something that doesn't affect Solr.If it must be a header, then you're going to need to write some custom code as suggesteed by Markus.Thanks,Shawn