Redis (2.6) on PowerPC ppc64

[Timo Schoeler]

Hi,

building an RPM of Redis for ppc64 is disabled in the .spec file:

ExcludeArch:      ppc64

https://github.com/antirez/redis/issues/400 <- there some reason can be found for this.

Became Redis clean to build on ppc64 with its 2.6 release? I didn't try yet. I have a bunch of Power machines running Fedora and would like to see it up and running again.

Cheers,

Timo

[Salvatore Sanfilippo]

Hello, in 2.6 I did many changes to make sure the release works with
big endian systems. Please could you try to build it and make test? It
should work without issues AFAIK.

Salvatore

> --
> You received this message because you are subscribed to the Google Groups
> "Redis DB" group.
> To view this discussion on the web visit
> https://groups.google.com/d/msg/redis-db/-/L2sCQOBizqwJ.
> 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 'antirez' Sanfilippo
open source developer - VMware
http://invece.org

Beauty is more important in computing than anywhere else in technology
because software is so complicated. Beauty is the ultimate defence
against complexity.
— David Gelernter

[Timo Schoeler]

Hi,

unfortunately, building fails (Fedora rawhide on ppc64; same env on x86_64 works):

[ok]: SLOWLOG - RESET subcommand works
[ok]: SLOWLOG - logged entry sanity check
[ok]: SLOWLOG - commands with too many arguments are trimmed
[ok]: SLOWLOG - too long arguments are trimmed
[25/32 done]: unit/slowlog (0 seconds)
Testing unit/scripting
[ok]: EVAL - Does Lua interpreter replies to our requests?
[ok]: EVAL - Lua integer -> Redis protocol type conversion
[ok]: EVAL - Lua string -> Redis protocol type conversion
[ok]: EVAL - Lua true boolean -> Redis protocol type conversion
[ok]: EVAL - Lua false boolean -> Redis protocol type conversion
[ok]: EVAL - Lua status code reply -> Redis protocol type conversion
[ok]: EVAL - Lua error reply -> Redis protocol type conversion
[ok]: EVAL - Lua table -> Redis protocol type conversion
[ok]: EVAL - Are the KEYS and ARGS arrays populated correctly?
[ok]: EVAL - is Lua able to call Redis API?

Logged warnings (pid 13925):
(none)

[exception]: Executing test client: NOSCRIPT No matching script. Please use EVAL..
NOSCRIPT No matching script. Please use EVAL.
    while executing
"::redis::redis_read_reply $fd"
    (procedure "::redis::__dispatch__" line 23)
    invoked from within
"[srv $level "client"] {*}$args"
    (procedure "r" line 7)
    invoked from within
"r evalsha 9bd632c7d33e571e9f24556ebed26c3479a87129 0"
    ("uplevel" body line 2)
    invoked from within
"uplevel 1 $code"
    (procedure "test" line 29)
    invoked from within
"test {EVALSHA - Can we call a SHA1 if already defined?} {
        r evalsha 9bd632c7d33e571e9f24556ebed26c3479a87129 0
    } {myval}"
    ("uplevel" body line 46)
    invoked from within
"uplevel 1 $code "
    (procedure "start_server" line 3)
    invoked from within
"start_server {tags {"scripting"}} {
    test {EVAL - Does Lua interpreter replies to our requests?} {
        r eval {return 'hello'} 0
    } {hello}
..."
    (file "tests/unit/scripting.tcl" line 1)
    invoked from within
"source $path"
    (procedure "execute_tests" line 4)
    invoked from within
"execute_tests $data"
    (procedure "test_client_main" line 9)
    invoked from within
"test_client_main $::test_server_port "
make[1]: *** [test] Error 1
make[1]: Leaving directory `/home/tis/rpmbuild/BUILD/redis-2.6.7/src'
make: *** [test] Error 2
error: Bad exit status from /var/tmp/rpm-tmp.UdU6Pl (%check)


RPM build errors:
    Bad exit status from /var/tmp/rpm-tmp.UdU6Pl (%check)

[Timo Schoeler]

(Tried for 2.6.7 and 2.6.6 -- same behaviour.)

[Salvatore Sanfilippo]

Thanks Timo, would it be possible for me to get an SSH access in order
to fix it and release a new Redis patch version ASAP?

Thank you in advance,
Salvatore

On Tue, Dec 11, 2012 at 9:22 AM, Timo Schoeler
<timothy....@googlemail.com> wrote:
> (Tried for 2.6.7 and 2.6.6 -- same behaviour.)
>

> --
> You received this message because you are subscribed to the Google Groups
> "Redis DB" group.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/redis-db/-/tg_nvuV-_KwJ.

[Timo Schoeler]

Please contact me privately timo at riscworks dot net

Cheers,

Timo

[Salvatore Sanfilippo]

Hello Timo,

I fixed the issue here:
https://github.com/antirez/redis/commit/705874e31db0196d8e1b23a28e8265dbdd2dc136

Thanks for your help.

Basically what was happening is that macros detecting the endianess
were not working properly on Linux/PPC64 so the code compiled to use
Little-endian specific issues.

Now that it's fixed everything works as expected and all the tests are
passed with success.

In the next hours I'll release Redis 2.6.8 with this fix, but you can
already use the latest commit in the 2.6 branch at github that is the
same code I'm going to release in 2.6.8.

Cheers,
Salvatore

> --
> You received this message because you are subscribed to the Google Groups
> "Redis DB" group.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/redis-db/-/K6LDThdEp8sJ.

[Timo Schoeler]

Hi Salvatore,

thanks a lot, I'll look at it ASAP.

Cheers,

Timo