mx/Jungfraujoch
2
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 72 Wiki Activity

Pre-flight: check the master file, not every data file of the run

Browse Source 
5f838bdb2 widened CheckOutputFilesAvailable from the master file to every data
file the run will write, so a stale _data_000001.h5 would be caught up front
rather than at the rename. The enumeration is unbounded: it is one
std::filesystem::exists per planned file, so a long run with a small
images_per_file - it can be set to 1 - is thousands of lookups, and they run
inside the 5 s PREFLIGHT ACK budget that BUSY heartbeats do not extend. On a
shared filesystem that turns a writable output into an "ACK timeout", which is
both wrong and unreadable. It also called HDF5Metadata::DataFileName for every
planned file, and that throws past file 999999 (9999 on SwissFEL), refusing a run
up front on a file number the acquisition would never reach.

Back to the master file alone, which is what the constructor checked before. A
data file in the way is caught per-writer at finalize again, as it was.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016L1qig74oYQzfUJJZbbxFh
This commit is contained in:
leonarski_f
2026-08-27 16:04:52 +02:00
co-authored by Claude Opus 5
parent 6dd36436b6
commit 547d2f5fa6
1 changed files with 10 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
writer/FileWriter.cpp
+10 -23
View File
@@ -222,37 +222,24 @@ void FileWriter::CheckOutputFilesAvailable(const StartMessage &msg, FileWriterFo
if (msg.overwrite.value_or(false))
return;
// Checked by the single writer that owns the master file (write_master_file - index 0 in a
// multi-writer TCP/ZMQ setup), and it checks the whole run, its siblings' data files included.
// Data files are staggered across writers by file number, so letting each writer check its own
// would race the siblings already creating them; one writer can speak for all of them because
// the master links the data files by relative name, so they share a directory by construction.
// Only the master file is checked, and only by the single writer that owns it (write_master_file
// - index 0 in a multi-writer TCP/ZMQ setup). Enumerating the data files instead would make one
// writer stat every file of the run: thousands of lookups for a long run with a small
// images_per_file, which on a network filesystem overruns the pre-flight's ACK budget and refuses
// an output that was perfectly writable. Those conflicts are caught per-writer at finalize.
if (!msg.write_master_file.value_or(false))
return;
auto refuse = [](const std::string &name) {
if (std::filesystem::exists(name))
throw JFJochException(JFJochExceptionCategory::FileWriteError,
"Output file already exists and overwrite is off: " + name);
};
const bool nxmx = format == FileWriterFormat::NXmxLegacy
|| format == FileWriterFormat::NXmxVDS
|| format == FileWriterFormat::NXmxIntegrated;
if (nxmx)
refuse(HDF5Metadata::MasterFileName(msg));
// NXmxIntegrated puts the images in the master file, so there are no data files to check.
if (format == FileWriterFormat::NXmxIntegrated || format == FileWriterFormat::NoFile)
if (!nxmx)
return;
const int64_t per_file = msg.images_per_file > 0
? msg.images_per_file
: static_cast<int64_t>(default_images_per_file);
const int64_t total_images = static_cast<int64_t>(msg.number_of_images);
for (int64_t file_number = 0; file_number * per_file < total_images; file_number++)
refuse(HDF5Metadata::DataFileName(msg, file_number));
const std::string name = HDF5Metadata::MasterFileName(msg);
if (std::filesystem::exists(name))
throw JFJochException(JFJochExceptionCategory::FileWriteError,
"Output file already exists and overwrite is off: " + name);
}
void FileWriter::Preflight(const StartMessage &request, bool trusted_path) {