Firebird lento

Vagner Costa

New Member
Mar 23, 2023
3
0
1
BRASIL
www.vcisoft.com.br
Olá pessoal
Instalei uma maquina virtual, nessas configurações!
1679543832433.png

1679544123456.png

1679544152223.png


Já está tudo rodando, todos os usuários conectados
entre 150 e 200 usuários!

Compartilhamento de arquivos está ok !

Porém o firebird versão 2.5 fica extremamente lento, e não consigo mudar de versão!
Já rodei todo o fórum procurando uma solução, pois instalando o mesmo banco de dados em uma maquina que não está virtualizada pelo proxmox, funciona normalmente sem travar

Alguém conseguiria me dar uma luz do que fazer para que o firebird pare de travar
 
You have two 6 core CPUs in the host, but given the VM 24 virtual cores.
This cannot run well.
The recommendation for other hypervisors (ESXi, HyperV) is to never give more resources to a VM than a physical CPU has.
If you want to use only 1 big VM I recommend not to configure more than 5 cores per socket and two sockets in the VM.
Your host should also be able to allocate CPU resources to perform e.g. network or storage tasks.

Tem duas CPUs de 6 núcleos no anfitrião, mas dado o VM 24 núcleos virtuais.
Isto não pode funcionar bem.
A recomendação para outros hipervisores (ESXi, HyperV) é nunca dar mais recursos a um VM do que um CPU físico tem.
Se quiser utilizar apenas 1 VM grande, recomendo não configurar mais de 5 núcleos por tomada e duas tomadas no VM.
O seu anfitrião deverá também ser capaz de atribuir recursos de CPU para executar, por exemplo, tarefas de rede ou de armazenamento.
 
Você tem duas CPUs de 6 núcleos no host, mas com 24 núcleos virtuais da VM.
Isso não pode correr bem.
A recomendação para outros hipervisores (ESXi, HyperV) é nunca fornecer mais recursos a uma VM do que uma CPU física.
Se você quiser usar apenas 1 VM grande, recomendo não configurar mais de 5 núcleos por soquete e dois soquetes na VM.
Seu host também deve ser capaz de alocar recursos de CPU para executar, por exemplo, tarefas de rede ou armazenamento.

Tem duas CPUs de 6 núcleos no host, mas dado o VM 24 núcleos virtuais.
Isto não pode funcionar bem.
Uma recomendação para outros hipervisores (ESXi, HyperV) é nunca dar mais recursos a um VM do que um CPU físico tem.
Se quiser utilizar apenas 1 VM grande, recomendo não configurar mais de 5 núcleos por tomada e duas tomadas no VM.
O seu anfitrião também deverá ser capaz de atribuir recursos de CPU para executar, por exemplo, tarefas de rede ou de processamento.
Certo
No meu caso aqui, teria como você me falar o que exatamente preciso alterar ?

essas são as configurações que está agora
1679588333903.png

E pelo o que entendi, você falou pra mudar para isso
1679588561151.png

estou certo?

tem algo mais?
 

Attachments

  • 1679588425918.png
    1679588425918.png
    41 KB · Views: 9
This is how it looks right.
This should make the VM work much smoother.

É assim que parece correcto.
Isto deve tornar o VM muito mais suave.
 
Boa Tarde,
@Vagner Costa

Alguma novidade sobre as alterações? O problema foi resolvido? Estou na mesma situação. Tenho uma maquina CT com Ubuntu 20.03 rodando o Firebird 2.5.9 e quando o usuário vai gravar os dados no sistema fica extremamente lento. Vi que os recursos de memória e processador não chegam a consumir 10% dos recursos reservados. Já verifiquei as permissões e também estão todas corretas.
 
Please post in English!
 
This is not a Proxmox Problem. Hardware and VM Sizing has the same restrictions on all Hypervisors.
 
I currently have a scenario where I have a VM that hosts the Rock Linux system, where my database is installed. I noticed that reading and writing the disk has a high latency. When configuring the same server with the same hardware outside of Proxmox I have much better disk reading and writing than when I have the server virtualized. I don't know if it's a limitation of Proxmox or if it's because it's a virtualized environment. Remembering that I did not configure any limitations for the disk.
 
Hi, what kind of disk is this and how did you set it up in Proxmox?
If it is a consumer SSD and you are using ZFS, you have a large overhead and the consumer SSDs cannot cope with the server workload, especially because they do not have a PLP (power loss protected cache).
 
SSD information:



=== START OF INFORMATION SECTION ===
Model Family: Phison Driven SSDs
Device Model: KINGSTON SA400S37240G
Serial Number: 50026B7784BD7659
LU WWN Device Id: 5 0026b7 784bd7659
Firmware Version: SBFKB1H5
User Capacity: 240,057,409,536 bytes [240 GB]
Sector Size: 512 bytes logical/physical
Rotation Rate: Solid State Device
TRIM Command: Available
Device is: In smartctl database 7.3/5319
ATA Version is: ACS-3 T13/2161-D revision 4
SATA Version is: SATA 3.2, 6.0 Gb/s (current: 6.0 Gb/s)
 
These are cheap consumer grade SSDs. With ZFS, you can expect 10k HDD Performance. When you would good Database Performance, please use Enterprise SSDs with PLP.
 
I am facing the same situation here. A VM with Firebird is extremely slow. I have tried everything, but nothing works. I am using SSDs in RAID 10, and both are datacenter versions. When migrating a VM to Hyper-V on a new server, it worked fine immediately.
 
I am facing the same situation here. A VM with Firebird is extremely slow. I have tried everything, but nothing works. I am using SSDs in RAID 10, and both are datacenter versions. When migrating a VM to Hyper-V on a new server, it worked fine immediately.
What does your VM configuration look like?
 
Uses a 40x Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz (2 Sockets) processor, a RAID driver is Perc H710p Mini 1GB with 4 Kingston DC600M SSDs in RAID10. Before, I was only using the indicated cores, but recently I changed and modified the sockets, and it was ZFS before as well. But I changed it! Both setups become slow when running systems with Firebird internally. 1718906204310.png
 
With a database of only 8 GB RAM, I personally find it a bit tight. If you really need 12 cores, the setup with NUMA is OK, but please check whether the application can handle NUMA. With 2x 10 core CPUs (40 threads) I would personally give a maximum of 10 cores per VM. It is best to only give as many cores as are absolutely necessary, as the CPU queue is then greatly relieved.

It is best to set the CPU type to host, which can also help a lot.
 
With a database of only 8 GB RAM, I personally find it a bit tight. If you really need 12 cores, the setup with NUMA is OK, but please check whether the application can handle NUMA. With 2x 10 core CPUs (40 threads) I would personally give a maximum of 10 cores per VM. It is best to only give as many cores as are absolutely necessary, as the CPU queue is then greatly relieved.

It is best to set the CPU type to host, which can also help a lot.
Thank you very much for the response. I am indeed concerned about whether NUMA is supported. I believe it might be better to disable it. I also have a question: is it better to assign 2 sockets, or should I just assign 10 cores to 1 socket by default? I recently adopted the idea of assigning 2 sockets and cores. But I believe Proxmox manages it automatically, just by assigning the cores. Is this idea correct? I also didn't know about choosing the host option. I have made the change here.
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!