The problem was caused by a bad ordering between the restart of the ORTE level tcp connections (in the OOB - out-of-band communication) and the Open MPI level tcp connections (BTLs). Before this commit ORTE would shutdown and restart the OOB completely before the OMPI level restarted its tcp connections. What would happen is that a socket descriptor used by the OMPI level on checkpoint was assigned to the ORTE level on restart. But the OMPI level had no knowledge that the socket descriptor it was previously using has been recycled so it closed it on restart. This caused the ORTE level to break as the newly created socket descriptor was closed without its knowledge.
The fix is to have the OMPI level shutdown tcp connections, allow the ORTE level to restart, and then allow the OMPi level to restart its connections. This seems obvious, and I'm surprised that this bug has not cropped up sooner. I'm confident that this specific problem has been fixed with this commit.
Thanks to Eric Roman and Tamer El Sayed for their help in identifying this problem, and patience while I was fixing it.
* Add a new state {{{OPAL_CRS_RESTART_PRE}}}. This state identifies when we are on the down slope of the INC (finalize-like) which is useful when you want to close, but not reopen a component set for fear of interfering with a lower level.
* Use this new state in OMPI level coordination. Here we want to make sure to play well with both the OMPI/BTL/TCP and ORTE/OOB/TCP components.
* Update ft_event functions in PML and BML to handle the new restart state.
* Add an additional flag to the error output in OOB/TCP so we can see what the socket descriptor was on failure as this can be helpful in debugging.
This commit was SVN r18276.
{{{
svn merge -r 18218:18240 https://svn.open-mpi.org/svn/ompi/tmp/jjh-scratch .
}}}
Contains:
* Primarily a fix for a user reported problem where a cached file descriptor is causing a SIGPIPE on restart.
* Cleanup some small memory leaks from using mca_base_param_env_var() - Thanks Jeff
* Cleanup ORTE FT tool compilation in non-FT builds - Thanks Tim P.
* Cleanup mpi interface with missplaced {{{OPAL_CR_ENTER_LIBRARY}}} - Thanks Terry
* Some other sundry cleanup items all dealing with C/R functionality in the trunk.
This commit was SVN r18241.
Like btl_tcp_disable_family, this parameter more or less disables
a whole address family. Though the sockets are still created, the
corresponding information isn't added to the connection strings.
Likewise, we don't try to connect to addresses matching the disabled
address family.
This is particularly important for multidomain clusters, where IPv4 is
oftenly filtered (firewalled), sometimes by simply dropping the packets
instead of rejecting them (thus causing a connection timeout instead of
a quick "no route to host").
This commit was SVN r18163.
Specifically, add two new APIs:
1. lost_route: allows the OOB to report that a connection has failed, thereby giving the routed module an opportunity to respond appropriately to its topology. Creating the API also allows each routed component to hold its own definition of "lifeline" - in some cases, this may be a single connection, but in others it may be multiple connections. Some modules may choose to re-route messaging if the lifeline or any other connection is lost, while others may choose to abort the job.
Both the tree and unity modules retain the current behavior and abort the job if the lifeline connection is lost, while ignoring other lost connections.
2. get_wireup_info: returns (in a provided buffer) info required to wireup connections for the specified job. Some routed modules do not need to return any info as they can wireup via alternative means, while some need to xchg data with their peers. If info is inserted into the buffer, the plm_base_launch_apps function will xcast the contents to the specified job.
The commit also removes the "lifeline" entry from the orte_process_info struct (and the associated ORTE_PROC_MY_LIFELINE definition) as the lifeline info is now contained within the respective routed module.
This commit was SVN r17969.
This has been a long-time problem. I tried to reduce the problem by having the orteds tell the HNP they were finalizing, and having the HNP wait until all orteds had reported or we timed out.
What was observed was that all the orteds were correctly reporting that they are leaving, but the HNP is able to exit before the orteds, thus closing the orteds lifeline socket and generating the error output. This is caused by the fact that the orteds have to whack all remaining session directories, which includes that blasted monster shared memory file! Cleaning up the SM file can take quite a while.
The HNP doesn't have that problem as there is no SM file there! So it gets out first.
What we had done in the past to resolve that problem was put a little test in the OOB that checks to see if we are finalizing. If we are, then we ignore the lifeline connection being lost. That check was still in the code - however, we had lost the line in orte_finalize that set the flag!!
This commit was SVN r17893.
This commit adds definition for a "lifeline" connection. For an HNP, there is no lifeline, so the lifeline proc is NULL. For a daemon, the lifeline is the HNP - the daemon should abort if it loses that connection.
For a proc using unity routed, the lifeline is the HNP since it connects directly to the HNP.
For a proc using tree routed, the lifeline is the local daemon.
Adjusted OOB to call abort if the lifeline (as opposed to HNP) connection is lost.
This commit was SVN r17761.
about linkers, have all OPAL, ORTE, and OMPI components '''not'' link
against the OPAL, ORTE, or OMPI libraries.
See ttp://www.open-mpi.org/community/lists/users/2007/10/4220.php for
details (or https://svn.open-mpi.org/trac/ompi/wiki/Linkers for a
better-formatted version of the same info).
This commit was SVN r16968.
has his own range which is defined by a min value and a range. By default
there is no limitation on the port range, which is exactly the same
behavior as before.
This commit was SVN r16584.
If we cannot resolve the route to the peer that we're trying to send
to, don't queue up the message in the TCP OOB -- instead, return it to
the upper layer (e.g., the RML) and let it decide what to do.
In the case of the routed RML, the tree component will queue it up for
later transmission. Hence, we don't want the message queued up both
here in the TCP OOB and the tree routed. Also see some more
discussion / explanation in #1171.
This commit was SVN r16540.
The following SVN revision numbers were found above:
r16513 --> open-mpi/ompi@7ae9589d70
The following Trac tickets were found above:
Ticket 1170 --> https://svn.open-mpi.org/trac/ompi/ticket/1170
The commit has been tested for C/R and Cray operations, and on Odin (SLURM, rsh) and RoadRunner (TM). I tried to update all environments, but obviously could not test them. I know that Windows needs some work, and have highlighted what is know to be needed in the odls process component.
This represents a lot of work by Brian, Tim P, Josh, and myself, with much advice from Jeff and others. For posterity, I have appended a copy of the email describing the work that was done:
As we have repeatedly noted, the modex operation in MPI_Init is the single greatest consumer of time during startup. To-date, we have executed that operation as an ORTE stage gate that held the process until a startup message containing all required modex (and OOB contact info - see #3 below) info could be sent to it. Each process would send its data to the HNP's registry, which assembled and sent the message when all processes had reported in.
In addition, ORTE had taken responsibility for monitoring process status as it progressed through a series of "stage gates". The process reported its status at each gate, and ORTE would then send a "release" message once all procs had reported in.
The incoming changes revamp these procedures in three ways:
1. eliminating the ORTE stage gate system and cleanly delineating responsibility between the OMPI and ORTE layers for MPI init/finalize. The modex stage gate (STG1) has been replaced by a collective operation in the modex itself that performs an allgather on the required modex info. The allgather is implemented using the orte_grpcomm framework since the BTL's are not active at that point. At the moment, the grpcomm framework only has a "basic" component analogous to OMPI's "basic" coll framework - I would recommend that the MPI team create additional, more advanced components to improve performance of this step.
The other stage gates have been replaced by orte_grpcomm barrier functions. We tried to use MPI barriers instead (since the BTL's are active at that point), but - as we discussed on the telecon - these are not currently true barriers so the job would hang when we fell through while messages were still in process. Note that the grpcomm barrier doesn't actually resolve that problem, but Brian has pointed out that we are unlikely to ever see it violated. Again, you might want to spend a little time on an advanced barrier algorithm as the one in "basic" is very simplistic.
Summarizing this change: ORTE no longer tracks process state nor has direct responsibility for synchronizing jobs. This is now done via collective operations within the MPI layer, albeit using ORTE collective communication services. I -strongly- urge the MPI team to implement advanced collective algorithms to improve the performance of this critical procedure.
2. reducing the volume of data exchanged during modex. Data in the modex consisted of the process name, the name of the node where that process is located (expressed as a string), plus a string representation of all contact info. The nodename was required in order for the modex to determine if the process was local or not - in addition, some people like to have it to print pretty error messages when a connection failed.
The size of this data has been reduced in three ways:
(a) reducing the size of the process name itself. The process name consisted of two 32-bit fields for the jobid and vpid. This is far larger than any current system, or system likely to exist in the near future, can support. Accordingly, the default size of these fields has been reduced to 16-bits, which means you can have 32k procs in each of 32k jobs. Since the daemons must have a vpid, and we require one daemon/node, this also restricts the default configuration to 32k nodes.
To support any future "mega-clusters", a configuration option --enable-jumbo-apps has been added. This option increases the jobid and vpid field sizes to 32-bits. Someday, if necessary, someone can add yet another option to increase them to 64-bits, I suppose.
(b) replacing the string nodename with an integer nodeid. Since we have one daemon/node, the nodeid corresponds to the local daemon's vpid. This replaces an often lengthy string with only 2 (or at most 4) bytes, a substantial reduction.
(c) when the mca param requesting that nodenames be sent to support pretty error messages, a second mca param is now used to request FQDN - otherwise, the domain name is stripped (by default) from the message to save space. If someone wants to combine those into a single param somehow (perhaps with an argument?), they are welcome to do so - I didn't want to alter what people are already using.
While these may seem like small savings, they actually amount to a significant impact when aggregated across the entire modex operation. Since every proc must receive the modex data regardless of the collective used to send it, just reducing the size of the process name removes nearly 400MBytes of communication from a 32k proc job (admittedly, much of this comm may occur in parallel). So it does add up pretty quickly.
3. routing RML messages to reduce connections. The default messaging system remains point-to-point - i.e., each proc opens a socket to every proc it communicates with and sends its messages directly. A new option uses the orteds as routers - i.e., each proc only opens a single socket to its local orted. All messages are sent from the proc to the orted, which forwards the message to the orted on the node where the intended recipient proc is located - that orted then forwards the message to its local proc (the recipient). This greatly reduces the connection storm we have encountered during startup.
It also has the benefit of removing the sharing of every proc's OOB contact with every other proc. The orted routing tables are populated during launch since every orted gets a map of where every proc is being placed. Each proc, therefore, only needs to know the contact info for its local daemon, which is passed in via the environment when the proc is fork/exec'd by the daemon. This alone removes ~50 bytes/process of communication that was in the current STG1 startup message - so for our 32k proc job, this saves us roughly 32k*50 = 1.6MBytes sent to 32k procs = 51GBytes of messaging.
Note that you can use the new routing method by specifying -mca routed tree - if you so desire. This mode will become the default at some point in the future.
There are a few minor additional changes in the commit that I'll just note in passing:
* propagation of command line mca params to the orteds - fixes ticket #1073. See note there for details.
* requiring of "finalize" prior to "exit" for MPI procs - fixes ticket #1144. See note there for details.
* cleanup of some stale header files
This commit was SVN r16364.
in all cases. This is now done in the oob tcp open function.
As a result, the unregistering have to be done in the close
function.
This commit was SVN r15603.
* General TCP cleanup for OPAL / ORTE
* Simplifying the OOB by moving much of the logic into the RML
* Allowing the OOB RML component to do routing of messages
* Adding a component framework for handling routing tables
* Moving the xcast functionality from the OOB base to its own framework
Includes merge from tmp/bwb-oob-rml-merge revisions:
r15506, r15507, r15508, r15510, r15511, r15512, r15513
This commit was SVN r15528.
The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
r15506
r15507
r15508
r15510
r15511
r15512
r15513
Cleanup ALL instances of output involving the printing of orte_process_name_t structures using the ORTE_NAME_ARGS macro so that the number of fields and type of data match. Replace those values with a new macro/function pair ORTE_NAME_PRINT that outputs a string (using the new thread safe data capability) so that any future changes to the printing of those structures can be accomplished with a change to a single point.
Note that I could not possibly find outputs that directly print the orte_process_name_t fields, but only dealt with those that used ORTE_NAME_ARGS. Hence, you may still have a few outputs that bark during compilation. Also, I could only verify those that fall within environments I can compile on, so other environments may yield some minor warnings.
This commit was SVN r15517.
Short description: major changes include -
1. singletons now fork/exec a local daemon to manage their operations.
2. the orte daemon code now resides in libopen-rte
3. daemons no longer use the orte triggering system during startup. Instead, they directly call back to their parent pls component to report ready to operate. A base function to count the callbacks has been provided.
I have modified all the pls components except xcpu and poe (don't understand either well enough to do it). Full functionality has been verified for rsh, SLURM, and TM systems. Compile has been verified for xgrid and gridengine.
This commit was SVN r15390.
VxWorks. Still some issues remaining, I'm sure.
Refs trac:1010
This commit was SVN r15320.
The following Trac tickets were found above:
Ticket 1010 --> https://svn.open-mpi.org/trac/ompi/ticket/1010
the multiple threads accessing the OOB/registry asynchronously via the
callbacks. The quickest solution (but definitively not the cleanest) is
to serialize these callbacks in such a way that at any given time
only one thread can execute a callbacks.
This commit was SVN r15086.
through the win dll using multiple threads, we have to insure that
the oob callbacks happens only in a synchronous way or really bad
things happens with the current design (blocking messages from a receive
callback).
This commit was SVN r15069.
single threaded builds. In its default configuration, all this does
is ensure that there's at least a good chance of threads building
based on non-threaded development (since the variable names will be
checked). There is also code to make sure that a "mutex" is never
"double locked" when using the conditional macro mutex operations.
This is off by default because there are a number of places in both
ORTE and OMPI where this alarm spews mega bytes of errors on a
simple test. So we have some work to do on our path towards
thread support.
Also removed the macro versions of the non-conditional thread locks,
as the only places they were used, the author of the code intended
to use the conditional thread locks. So now you have upper-case
macros for conditional thread locks and lowercase functions for
non-conditional locks. Simple, right? :).
This commit was SVN r15011.
structures in the system. Instead of using memcmp, use the ns function.
This won't cause a problem as long as all three elements of the name are
ints, but if they have different sizes, alignment and padding rules
can cause memcmp() to compare padding space, which rarely holds a sane
value.
This commit was SVN r14998.
