Discussion:
[VOTE] Release Apache Traffic Server 7.1.0 (RC1)
Leif Hedstrom
2017-07-18 08:35:32 UTC
Permalink
I've prepared a release for 7.1.0 (RC1) which is the next major version of Apache Traffic Server. As per our new release schedule and process, v7.1.x is an Long-Term Support (LTS) release. This is detailed in our Release Management document:

https://cwiki.apache.org/confluence/display/TS/Release+Management


Release notes for 7.1.0:

https://github.com/apache/trafficserver/milestone/2?closed=1


This release of v7.1.0 is backwards compatible with v7.0.0, for some details as to what’s in v.7.1.x see

https://cwiki.apache.org/confluence/display/TS/What%27s+New+in+v7.1.x


Information about upgrading to this release from previous major versions is available at:

https://cwiki.apache.org/confluence/display/TS/Upgrading+to+v7.0

The artifacts are available for download at:

http://people.apache.org/~zwoop/rel-candidates/


Checksums:

MD5: 3732ae52cc8a2612de9a27a6fb67c50a *trafficserver-7.1.0-rc1.tar.bz2
SHA1: a4ea9bff7be4d99e4beb9208603dd2e6599d7304 *trafficserver-7.1.0-rc1.tar.bz2


This corresponds to git:

Hash: 367439f813a0c28f2015522a34634c7dc1ad3831
Tag: 7.1.0-rc1


Which can be verified with the following command:

$ git tag -v 7.1.0-rc1


All code signing keys are available here:

https://dist.apache.org/repos/dist/dev/trafficserver/KEYS

Make sure you refresh from a key server to get all relevant signatures.

This vote is open until July 21th (yes, a little bit short, but the changes from RC0 to RC1 as very small). Please vote soon and often!

Cheers,

— Leif
Leif Hedstrom
2017-07-18 11:52:38 UTC
Permalink
Post by Leif Hedstrom
https://cwiki.apache.org/confluence/display/TS/Release+Management
+1 from me, tested on CentOS6.

— Leif
Reindl Harald
2017-07-18 15:29:28 UTC
Permalink
Post by Leif Hedstrom
I've prepared a release for 7.1.0 (RC1) which is the next major version of Apache Traffic Server. As per our new release schedule and process, v7.1.x is an Long-Term Support (LTS) release. This is detailed in our Release
looks good so far on Fedora 25 x86_64

but the still ongoing complaining about a readonly /etc starts to annoy
me *really* because it also prevents ATS from a soft config reload and a
hard restart for just add a new remap-pair is just bad

below also some compiler warnings (LTO build)
_______________________________________________________________________

[Jul 18 17:25:00.729] Manager {0x7f12b4f2f940} NOTE:
[Rollback::openFile] Open of log_hosts.config failed: Read-only file system
_______________________________________________________________________

SSLCertLookup.cc:340:1: warning: 'char* reverse_dns_name(const char*,
char (&)[257])' defined but not used [-Wunused-function]
reverse_dns_name(const char *hostname, char
(&reversed)[TS_MAX_HOST_NAME_LEN + 1])
^~~~~~~~~~~~~~~~
In file included from SSLCertLookup.cc:34:0:
../../lib/ts/TestBox.h:59:1: warning: 'bool
{anonymous}::TestBox::check(bool, const char*, ...)' defined but not
used [-Wunused-function]
TestBox::check(bool result, char const *fmt, ...)
^~~~~~~
HdrTSOnly.cc: In member function 'HTTPHdr::parse_req(HTTPParser*,
IOBufferReader*, int*, bool, bool)':
HdrHeap.h:226:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_ref_count_ptr = m_ronly_heap[i].m_ref_count_ptr;
~~~~~~~~~~~~~~^
HdrHeap.h:227:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_heap_start = m_ronly_heap[i].m_heap_start;
~~~~~~~~~~~~~~^
HdrHeap.h:228:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_heap_len = m_ronly_heap[i].m_heap_len;
~~~~~~~~~~~~~~^
HdrHeap.h:229:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_locked = m_ronly_heap[i].m_locked;
~~~~~~~~~~~~~~^
HdrHeap.h:230:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_ref_count_ptr = NULL;
~~~~~~~~~~~~~~^
HdrHeap.h:231:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_heap_start = NULL;
~~~~~~~~~~~~~~^
HdrHeap.h:232:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_heap_len = 0;
~~~~~~~~~~~~~~^
HdrHeap.h:233:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_locked = false;
~~~~~~~~~~~~~~^
HdrHeap.h:230:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_ref_count_ptr = NULL;
~~~~~~~~~~~~~~^
HdrTSOnly.cc: In member function 'HTTPHdr::parse_resp(HTTPParser*,
IOBufferReader*, int*, bool)':
HdrHeap.h:226:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_ref_count_ptr = m_ronly_heap[i].m_ref_count_ptr;
~~~~~~~~~~~~~~^
HdrHeap.h:227:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_heap_start = m_ronly_heap[i].m_heap_start;
~~~~~~~~~~~~~~^
HdrHeap.h:228:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_heap_len = m_ronly_heap[i].m_heap_len;
~~~~~~~~~~~~~~^
HdrHeap.h:229:57: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[j].m_locked = m_ronly_heap[i].m_locked;
~~~~~~~~~~~~~~^
HdrHeap.h:230:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_ref_count_ptr = NULL;
~~~~~~~~~~~~~~^
HdrHeap.h:231:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_heap_start = NULL;
~~~~~~~~~~~~~~^
HdrHeap.h:232:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_heap_len = 0;
~~~~~~~~~~~~~~^
HdrHeap.h:233:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_locked = false;
~~~~~~~~~~~~~~^
HdrHeap.h:230:23: warning: array subscript is above array bounds
[-Warray-bounds]
m_ronly_heap[i].m_ref_count_ptr = NULL;
~~~~~~~~~~~~~~^
In file included from ProxyConfig.cc:27:0:
../lib/ts/TestBox.h:59:1: warning: 'bool
{anonymous}::TestBox::check(bool, const char*, ...)' defined but not
used [-Wunused-function]
TestBox::check(bool result, char const *fmt, ...)
^~~~~~~
In file included from LogObject.cc:39:0:
../../lib/ts/TestBox.h:59:1: warning: 'bool
{anonymous}::TestBox::check(bool, const char*, ...)' defined but not
used [-Wunused-function]
TestBox::check(bool result, char const *fmt, ...)
^~~~~~~
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:51:8:
warning: type 'struct Machine' violates the C++ One Definition Rule [-Wodr]
struct Machine {
^
UglyLogStubs.cc:68:8: note: a different type is defined in another
translation unit
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:54:9:
note: the first difference of corresponding definitions is field 'hostname'
char *hostname; // name of the internet host
^
UglyLogStubs.cc:68:8: note: a type with different number of fields is
defined in another translation unit
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:81:16:
warning: type of 'instance' does not match original declaration
[-Wlto-type-mismatch]
static self *instance();
^
UglyLogStubs.cc:72:1: note: 'instance' was previously declared here
UglyLogStubs.cc:72:1: note: code may be misoptimized unless
-fno-strict-aliasing is used
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:51:8:
warning: type 'struct Machine' violates the C++ One Definition Rule [-Wodr]
struct Machine {
^
UglyLogStubs.cc:68:8: note: a different type is defined in another
translation unit
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:54:9:
note: the first difference of corresponding definitions is field 'hostname'
char *hostname; // name of the internet host
^
UglyLogStubs.cc:68:8: note: a type with different number of fields is
defined in another translation unit
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:81:16:
warning: type of 'instance' does not match original declaration
[-Wlto-type-mismatch]
static self *instance();
UglyLogStubs.cc:72:1: note: 'instance' was previously declared here
UglyLogStubs.cc:72:1: note: code may be misoptimized unless
-fno-strict-aliasing is used
UglyLogStubs.cc:68:8: note: a type with different number of fields is
defined in another translation unit
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:81:16:
warning: type of 'instance' does not match original declaration
[-Wlto-type-mismatch]
static self *instance();
^
UglyLogStubs.cc:72:1: note: 'instance' was previously declared here
UglyLogStubs.cc:72:1: note: code may be misoptimized unless
-fno-strict-aliasing is used
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:51:8:
warning: type 'struct Machine' violates the C++ One Definition Rule [-Wodr]
struct Machine {
^
UglyLogStubs.cc:68:8: note: a different type is defined in another
translation unit
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:54:9:
note: the first difference of corresponding definitions is field 'hostname'
char *hostname; // name of the internet host
^
UglyLogStubs.cc:68:8: note: a type with different number of fields is
defined in another translation unit
/home/builduser/rpmbuild/BUILD/trafficserver-7.1.0/iocore/utils/I_Machine.h:81:16:
warning: type of 'instance' does not match original declaration
[-Wlto-type-mismatch]
static self *instance();
^
UglyLogStubs.cc:72:1: note: 'instance' was previously declared here
UglyLogStubs.cc:72:1: note: code may be misoptimized unless
-fno-strict-aliasing is used
ClusterHandlerBase.cc:35:17: warning: type of 'cluster_packet_tos' does
not match original declaration [-Wlto-type-mismatch]
ClusterProcessor.cc:377:15: note: type 'long unsigned int' should match
type 'uint32_t'
/usr/include/stdint.h:51:23: note: the incompatible type is defined here
typedef unsigned int uint32_t;
^
ClusterProcessor.cc:377:15: note: 'cluster_packet_tos' was previously
declared here
ClusterProcessor.cc:377:15: note: code may be misoptimized unless
-fno-strict-aliasing is used
ClusterHandlerBase.cc:34:17: warning: type of 'cluster_packet_mark' does
not match original declaration [-Wlto-type-mismatch]
ClusterProcessor.cc:376:15: note: type 'long unsigned int' should match
type 'uint32_t'
/usr/include/stdint.h:51:23: note: the incompatible type is defined here
typedef unsigned int uint32_t;
^
ClusterProcessor.cc:376:15: note: 'cluster_packet_mark' was previously
declared here
ClusterProcessor.cc:376:15: note: code may be misoptimized unless
-fno-strict-aliasing is used
ClusterHandlerBase.cc:33:17: warning: type of 'cluster_sockopt_flags'
does not match original declaration [-Wlto-type-mismatch]
ClusterProcessor.cc:375:15: note: type 'long unsigned int' should match
type 'uint32_t'
/usr/include/stdint.h:51:23: note: the incompatible type is defined here
typedef unsigned int uint32_t;
^
ClusterProcessor.cc:375:15: note: 'cluster_sockopt_flags' was previously
declared here
ClusterProcessor.cc:375:15: note: code may be misoptimized unless
-fno-strict-aliasing is used
HttpSM.cc: In member function 'do_redirect':
HttpSM.cc:7836:47: warning: 'redir_len' may be used uninitialized in
this function [-Wmaybe-uninitialized]
HttpSM.cc:7834:13: note: 'redir_len' was declared here
HttpTransact.cc: In function 'issue_revalidate':
HttpTransact.cc:2329:19: warning: 'length' may be used uninitialized in
this function [-Wmaybe-uninitialized]
HttpTransact.cc:2325:11: note: 'length' was declared here
HttpTransact.cc:2318:114: warning: 'length' may be used uninitialized in
this function [-Wmaybe-uninitialized]
HttpTransact.cc:2315:11: note: 'length' was declared here
Phillip Moore
2017-07-19 20:37:18 UTC
Permalink
+1

Have running on internal staging and production traffic and it is working
fine. I had no build troubles on SL6.9.

--pdm
Susan Hinrichs
2017-07-20 14:41:13 UTC
Permalink
+1 Running for 24 hours on a production box without problems.

On Thursday, July 20, 2017, 1:19:26 AM CDT, Phillip Moore <***@pobox.com> wrote:

+1

Have running on internal staging and production traffic and it is working
fine.  I had no build troubles on SL6.9.

--pdm
Bryan Call
2017-07-20 14:43:24 UTC
Permalink
+1 - Running in production on RHEL 6.8

-Bryan
Post by Leif Hedstrom
https://cwiki.apache.org/confluence/display/TS/Release+Management
https://github.com/apache/trafficserver/milestone/2?closed=1
This release of v7.1.0 is backwards compatible with v7.0.0, for some details as to what’s in v.7.1.x see
https://cwiki.apache.org/confluence/display/TS/What%27s+New+in+v7.1.x
https://cwiki.apache.org/confluence/display/TS/Upgrading+to+v7.0
http://people.apache.org/~zwoop/rel-candidates/
MD5: 3732ae52cc8a2612de9a27a6fb67c50a *trafficserver-7.1.0-rc1.tar.bz2
SHA1: a4ea9bff7be4d99e4beb9208603dd2e6599d7304 *trafficserver-7.1.0-rc1.tar.bz2
Hash: 367439f813a0c28f2015522a34634c7dc1ad3831
Tag: 7.1.0-rc1
$ git tag -v 7.1.0-rc1
https://dist.apache.org/repos/dist/dev/trafficserver/KEYS
Make sure you refresh from a key server to get all relevant signatures.
This vote is open until July 21th (yes, a little bit short, but the changes from RC0 to RC1 as very small). Please vote soon and often!
Cheers,
— Leif
Phillip Moore
2017-07-21 00:51:33 UTC
Permalink
Anyone experience high CPU with 7.1.0 rc1 ? I have it running on 2 nodes
taking production load and it is doing comparable work to other 6.2.1 nodes
in both cases. in one case the CPU climbed to 100% of all 32 cores and
stayed stuck there, in the other one the CPU is in line with other nodes.
I'm not sure what the difference is or cause. The work load is pretty
similar.

perf top reported it in spin_lock.

--pdm
Post by Leif Hedstrom
Post by Leif Hedstrom
I've prepared a release for 7.1.0 (RC1) which is the next major version
of Apache Traffic Server. As per our new release schedule and process,
v7.1.x is an Long-Term Support (LTS) release. This is detailed in our
Post by Leif Hedstrom
https://cwiki.apache.org/confluence/display/TS/Release+Management
https://github.com/apache/trafficserver/milestone/2?closed=1
This release of v7.1.0 is backwards compatible with v7.0.0, for some
details as to what’s in v.7.1.x see
Post by Leif Hedstrom
https://cwiki.apache.org/confluence/display/TS/What%
27s+New+in+v7.1.x
Post by Leif Hedstrom
Information about upgrading to this release from previous major versions
https://cwiki.apache.org/confluence/display/TS/Upgrading+to+v7.0
http://people.apache.org/~zwoop/rel-candidates/
MD5: 3732ae52cc8a2612de9a27a6fb67c50a
*trafficserver-7.1.0-rc1.tar.bz2
Post by Leif Hedstrom
SHA1: a4ea9bff7be4d99e4beb9208603dd2e6599d7304
*trafficserver-7.1.0-rc1.tar.bz2
Post by Leif Hedstrom
Hash: 367439f813a0c28f2015522a34634c7dc1ad3831
Tag: 7.1.0-rc1
$ git tag -v 7.1.0-rc1
https://dist.apache.org/repos/dist/dev/trafficserver/KEYS
Make sure you refresh from a key server to get all relevant signatures.
This vote is open until July 21th (yes, a little bit short, but the
changes from RC0 to RC1 as very small). Please vote soon and often!
Post by Leif Hedstrom
Cheers,
— Leif
Bryan Call
2017-07-21 07:26:27 UTC
Permalink
Can you run sudo perf top -g, do a search (“/“) for spin_lock, and see what is calling it?

-Bryan
Anyone experience high CPU with 7.1.0 rc1 ? I have it running on 2 nodes taking production load and it is doing comparable work to other 6.2.1 nodes in both cases. in one case the CPU climbed to 100% of all 32 cores and stayed stuck there, in the other one the CPU is in line with other nodes. I'm not sure what the difference is or cause. The work load is pretty similar.
perf top reported it in spin_lock.
--pdm
https://cwiki.apache.org/confluence/display/TS/Release+Management <https://cwiki.apache.org/confluence/display/TS/Release+Management>
https://github.com/apache/trafficserver/milestone/2?closed=1 <https://github.com/apache/trafficserver/milestone/2?closed=1>
This release of v7.1.0 is backwards compatible with v7.0.0, for some details as to what’s in v.7.1.x see
https://cwiki.apache.org/confluence/display/TS/What%27s+New+in+v7.1.x <https://cwiki.apache.org/confluence/display/TS/What%27s+New+in+v7.1.x>
https://cwiki.apache.org/confluence/display/TS/Upgrading+to+v7.0 <https://cwiki.apache.org/confluence/display/TS/Upgrading+to+v7.0>
http://people.apache.org/~zwoop/rel-candidates/ <http://people.apache.org/~zwoop/rel-candidates/>
MD5: 3732ae52cc8a2612de9a27a6fb67c50a *trafficserver-7.1.0-rc1.tar.bz2
SHA1: a4ea9bff7be4d99e4beb9208603dd2e6599d7304 *trafficserver-7.1.0-rc1.tar.bz2
Hash: 367439f813a0c28f2015522a34634c7dc1ad3831
Tag: 7.1.0-rc1
$ git tag -v 7.1.0-rc1
https://dist.apache.org/repos/dist/dev/trafficserver/KEYS <https://dist.apache.org/repos/dist/dev/trafficserver/KEYS>
Make sure you refresh from a key server to get all relevant signatures.
This vote is open until July 21th (yes, a little bit short, but the changes from RC0 to RC1 as very small). Please vote soon and often!
Cheers,
— Leif
Leif Hedstrom
2017-07-21 09:41:37 UTC
Permalink
Can you run sudo perf top -g, do a search (“/“) for spin_lock, and see what is calling it?
-Bryan
Anyone experience high CPU with 7.1.0 rc1 ? I have it running on 2 nodes taking production load and it is doing comparable work to other 6.2.1 nodes in both cases. in one case the CPU climbed to 100% of all 32 cores and stayed stuck there, in the other one the CPU is in line with other nodes. I'm not sure what the difference is or cause. The work load is pretty similar.
I checked all my boxes (we have quite a few deployed with 7.1.0 now), and none of them is showing this behavior (so far).

— Leif
Leif Hedstrom
2017-07-23 23:45:38 UTC
Permalink
Post by Leif Hedstrom
https://cwiki.apache.org/confluence/display/TS/Release+Management
I’m going to call this, with 5 +1 votes (3 binding), and no -1’s.

I will prepare the release announcements etc. tomorrow.

— Leif

Loading...