| From: | mahamood hussain <hussain(dot)ieg(at)gmail(dot)com> |
|---|---|
| To: | Licio Matos <licio(dot)matos(at)gmail(dot)com> |
| Cc: | Ireneusz Pluta <ipluta(at)wp(dot)pl>, pgsql-admin(at)lists(dot)postgresql(dot)org |
| Subject: | Re: Urgent !!!! Tables inaccessible postgres v17.6 |
| Date: | 2026-08-14 09:10:18 |
| Message-ID: | CAGc_7H=-oXr9t_45dQzNyW99fz78HEkXpF-_Zww41wAJ6247iQ@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-admin |
Hi Matos,
Thanks for the response.
The PostgreSQL data volume is deployed using LVM striping, not mdadm. It
uses four Azure NVMe disks, combined into a ~2 TB LVM logical volume. The
LVM stripe size is 128 KiB.
I also looked into PostgreSQL tablespaces. While PostgreSQL supports
tablespaces, they don't provide the same capability as DB2 storage groups
and storage paths.
In DB2, a tablespace can be associated with a storage group, and the
storage group can have multiple storage paths. This allows the tablespace's
I/O to be distributed across multiple disks:
DB2 Database
|
Tablespace
|
Storage Group
|
+-----------+-----------+
| | |
Path 1 Path 2 Path 3
| | |
Disk 1 Disk 2 Disk 3
This is different from PostgreSQL, where a tablespace points to a single
filesystem/location. PostgreSQL itself doesn't provide the same
storage-group concept where one tablespace can directly use multiple
independent storage paths.
Therefore, if we want to distribute the I/O for a PostgreSQL tablespace
across multiple disks, that distribution needs to happen at the OS/storage
layer, for example through LVM striping or RAID.
This was the main reason I considered LVM striping for the PostgreSQL data
volume — to achieve a similar I/O distribution model to what we had with
DB2 storage groups and storage paths.
Please let me know if there is a better PostgreSQL/Azure architecture for
achieving the same level of IOPS and throughput without relying on OS-level
striping.
On Thu, Aug 13, 2026 at 5:49 PM Licio Matos <licio(dot)matos(at)gmail(dot)com> wrote:
> Hi,
>
> Just to confirm, are you using 4 disks in raid 0 from azure ssd v2? How
> they are deployed ,LVM or mdadm? What’s is the page size of the stripe?
>
> About tablespace, you can use tablespaces in PostgreSQL the same way.
>
> https://www.postgresql.org/docs/current/manage-ag-tablespaces.html
>
>
> Licio Matos
>
> Em qui., 13 de ago. de 2026 às 06:58, Ireneusz Pluta <ipluta(at)wp(dot)pl>
> escreveu:
>
>> W dniu 13.08.2026 o 10:55 AM, mahamood hussain pisze:
>> > are there any known limitations or risks with using RAID 0 even with
>> > the underlying durability provided by Premium SSD v2?
>> using RAID 0 is always a risk, regardless how much premium your hardware
>> is. Use RAID10 if you value your data.
>>
>>
>>
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Laurenz Albe | 2026-08-14 22:09:11 | Re: Urgent !!!! Tables inaccessible postgres v17.6 |
| Previous Message | Michael Banck | 2026-08-13 12:37:20 | Re: pg_upgrade from 17.10 to 18.4 on Ubuntu fails: Finding the real data directory for the target cluster sh: 1: (null)/postgres: not found |