2.2 going stable in a few days. Comments?

[Salvatore Sanfilippo]

Hello folks!

I think we should push 2.2 as stable ASAP. It's already used by many
high profile redis users with success, and works great.
So I'm asking here for comments by users with 2.2 in production, if
everything is working well.

I'm especially interested in comments about:

1) new maxmemory policies and how this is working.
2) replication.
3) systems with VM enabled.

But all the comments are appreciated.

Thank you!
Salvatore

--
Salvatore 'antirez' Sanfilippo
open source developer - VMware

http://invece.org
"We are what we repeatedly do. Excellence, therefore, is not an act,
but a habit." -- Aristotele

[Jeremy Zawodny]

Salvatore,

We're using:

maxmemory 6gb

maxmemory-policy volatile-lru

maxmemory-samples 3

And it's working quite well.  Someone even accidentally shoved way more data into redis than expected while I was on vacation and nobody really noticed because redis behaved just as expected.  We're using replication and have seen no problems there as well.

No VM in our environment.

This is all on the same 2.1.7 code that we've been running for a few weeks now:

uptime_in_days:62

Jeremy

--
You received this message because you are subscribed to the Google Groups "Redis DB" group.
To post to this group, send email to redi...@googlegroups.com.
To unsubscribe from this group, send email to redis-db+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/redis-db?hl=en.

[Salvatore Sanfilippo]

On Thu, Jan 13, 2011 at 5:04 PM, Jeremy Zawodny <Jer...@zawodny.com> wrote:
> Salvatore,
> We're using:
> maxmemory 6gb
> maxmemory-policy volatile-lru
> maxmemory-samples 3
> And it's working quite well.  Someone even accidentally shoved way more data

Thank you Jeremy, very cool.

Please could you report the RSS usage of this box? In 2.2 it is also
listed in INFO as you probably already noticed. Thank you.

Cheers,

[Jeremy Zawodny]

Here's the whole INFO output from it.  (There are 4 procs per box, this is just one of them.)

redis_version:2.1.7

redis_git_sha1:00000000

redis_git_dirty:0

arch_bits:64

multiplexing_api:epoll

process_id:2194

uptime_in_seconds:5420659

uptime_in_days:62

lru_clock:1567298

used_cpu_sys:15994.72

used_cpu_user:25930.20

used_cpu_sys_childrens:130.96

used_cpu_user_childrens:56.52

connected_clients:972

connected_slaves:1

blocked_clients:0

used_memory:5531826736

used_memory_human:5.15G

used_memory_rss:6413373440

mem_fragmentation_ratio:1.16

use_tcmalloc:0

loading:0

aof_enabled:0

changes_since_last_save:220874892

bgsave_in_progress:0

last_save_time:1289596978

bgrewriteaof_in_progress:0

total_connections_received:31528365

total_commands_processed:561548640

expired_keys:16551546

keyspace_hits:305495171

keyspace_misses:91977432

hash_max_zipmap_entries:512

hash_max_zipmap_value:512

pubsub_channels:0

pubsub_patterns:0

vm_enabled:0

role:master

db0:keys=4512733,expires=4512730

[Salvatore Sanfilippo]

Awesome, so low fragmentation after all this commands processed and
with a big dataset looks very encouraging.

Thank you very much,
Salvatore

[Jeremy Zawodny]

Yes, we're very happy with it.

Virtually all our keys have an expire time (as you can tell).  And I'd say that easily 75% of our keys are ZSETs (many are very small).  The remaining keys are mostly simple key/value pairs.  There are a few lists in there too (a simple queue implementation.)

Jeremy

[Александр Лозовюк]

We have not so large installation - 7 instance (servers, small/medium
EC2 instance) with master-slave replication, less then 1Gb data, but
active used Pub/Sub for dustribute data acros server at real-time
(e.g. at 5 000 msg/sec at master in peack, each msg is JSON 500 - 3000
bytes length). Its work fine - all server 2.0.1/32 bit, one 2.1.8,
uptime 92 days.

2011/1/13 Jeremy Zawodny <Jer...@zawodny.com>:

--
C уважением, Александр Лозовюк
Alpha-Beta-Release Blog
http://abrdev.com

[Genlok]

Submitted an issue on redis.io, As you wanted to declare redis 2.1.10
stable, I writing
it here too. As 2.1.10 segfaulted on our system. Please see the issue.

Best regards,
Patric

[Salvatore Sanfilippo]

Hello!

is this issue 434? Thanks a ton for reporting. This is definitely a
real problem and probably not hard to fix.
We are working on it and will be fixed in short time.

This is the first crash we get in 2.2, it is due to an interaction
between blocking pop and closing the connection.

Cheers,
Salvatore

> --
> You received this message because you are subscribed to the Google Groups "Redis DB" group.
> To post to this group, send email to redi...@googlegroups.com.
> To unsubscribe from this group, send email to redis-db+u...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/redis-db?hl=en.
>
>

--

[Salvatore Sanfilippo]

Update: the bug was fixed by Pieter both in the 2.2 and unstable
branch after 29 minutes it was submitted ;)

The bug did not effected 2.0 btw.

Cheers,
Salvatore