[
https://issues.apache.org/jira/browse/TS-3611?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14549933#comment-14549933
]
Nikolai Gorchilov edited comment on TS-3611 at 5/20/15 4:24 AM:
----------------------------------------------------------------
Unfortunately all of our production servers run on the same binary build with
interim caching enabled.
We're aware that the interim storage will dissapear in v6 and we're planning to
migrate to bcache later this year. In relation to that - are there any
whitepapers/case studies regarding an optimal setup of bcache with ATS - like
SSD to rotational disk size ratios, recommended bcache mode of operation,
fine-tunings, etc. I couldn't find any relevant information.
was (Author: ngorchilov):
Unfortunately all of our productions servers run on the same binary build with
interim caching enabled.
We're aware that the interim storage will dissapear in v6 and we're planning to
migrate to bcache later this year. In relation to that - are there any
whitepapers/case studies regarding an optimal setup of bcache with ATS - like
SSD to rotational disk size ratios, recommended bcache mode of operation,
fine-tunings, etc. I couldn't find any relevant information.
> Segmentation fault: Address not mapped to object
> ------------------------------------------------
>
> Key: TS-3611
> URL: https://issues.apache.org/jira/browse/TS-3611
> Project: Traffic Server
> Issue Type: Bug
> Affects Versions: 5.3.0
> Reporter: Nikolai Gorchilov
> Labels: crash
> Fix For: 6.0.0
>
>
> Experienced the following crash while running ATS 5.3.0 on custom Linux
> distro:
> {noformat}
> traffic_server: Segmentation fault (Address not mapped to object [0x10])
> traffic_server - STACK TRACE:
> /z/bin/traffic_server(crash_logger_invoke(int, siginfo_t*,
> void*)+0x8e)[0x4ad5fe]
> /lib64/libpthread.so.0(+0x105f0)[0x2b63a09b55f0]
> /z/bin/traffic_server(HostDBProcessor::getbyname_imm(Continuation*, void
> (Continuation::*)(HostDBInfo*), char const*, int, HostDBProcessor::Options
> const&)+0x72)[0x68c652]
> /z/bin/traffic_server(HttpSM::do_hostdb_lookup()+0xb8)[0x5b2f48]
> /z/bin/traffic_server(HttpSM::set_next_state()+0xdb5)[0x5c1815]
> /z/bin/traffic_server(HttpSM::state_send_server_request_header(int,
> void*)+0x15a)[0x5bd5da]
> /z/bin/traffic_server(HttpSM::main_handler(int, void*)+0xbd)[0x5c282d]
> /z/bin/traffic_server[0x749be6]
> /z/bin/traffic_server(NetHandler::mainNetEvent(int, Event*)+0x1f2)[0x73c742]
> /z/bin/traffic_server(EThread::execute()+0xa4a)[0x76ad4a]
> /z/bin/traffic_server[0x7699ba]
> /lib64/libpthread.so.0(+0x7294)[0x2b63a09ac294]
> /lib64/libc.so.6(clone+0x6d)[0x2b63a16d471d]
> {noformat}
> UPDATE: Few hours later, another server, same crash reason, different
> call-stack:
> {noformat}
> traffic_server: Segmentation fault (Address not mapped to object
> [0x2adc4759bff0])
> traffic_server - STACK TRACE:
> /z/bin/traffic_server(crash_logger_invoke(int, siginfo*,
> void*)+0x99)[0x4b0e49]
> /lib64/libpthread.so.0(+0x10630)[0x2ad84e658630]
> /lib64/libc.so.6(+0x903de)[0x2ad84f5273de]
> /z/bin/traffic_server(Vol::aggWrite(int, void*)+0x715)[0x6ff6a5]
> /z/bin/traffic_server(Vol::evacuateDocReadDone(int, Event*)+0x3fa)[0x70092a]
> /z/bin/traffic_server(AIOCallbackInternal::io_complete(int,
> void*)+0x3d)[0x6812fd]
> /z/bin/traffic_server(EThread::process_event(Event*, int)+0x10f)[0x75e77f]
> /z/bin/traffic_server(EThread::execute()+0x6ef)[0x75f28f]
> /z/bin/traffic_server[0x75db0a]
> /lib64/libpthread.so.0(+0x72d4)[0x2ad84e64f2d4]
> /lib64/libc.so.6(clone+0x6d)[0x2ad84f57e74d]
> {noformat}
> Here're the relevant configure options of my build:
> {noformat}
> ./configure \
> --with-xml=libxml2 \
> --disable-static \
> --disable-spdy \
> --enable-interim-cache \
> --enable-tproxy \
> --enable-hwloc \
> --enable-experimental-plugins \
> --enable-example-plugins
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
