BitBOINC

It mentioned about disk usage limit exceeded, but I have around 40GB of disk space allowed for BOINC, while peak disk usage is just 12.02 MB. Is there any reason as to what happened

<core_client_version>8.2.11</core_client_version>
<![CDATA[
<message>
Disk usage limit exceeded</message>
<stderr_txt>
Keraunos GPU (bounded discrete-log search, CUDA) v1
bitboinc -- https://bitboinc.athena.org.tr/
coded by Alperen Yavuz

</stderr_txt>
]]>

All workunits for Android fail for Keraunos (bounded discrete-log search) v0.14

<core_client_version>8.2.11</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)</message>
<stderr_txt>
Keraunos OpenCL (bounded discrete-log search, OpenCL) v1
bitboinc -- https://bitboinc.athena.org.tr/
coded by Alperen Yavuz
cl_dynload_init: no OpenCL ICD found on this host (tried libOpenCL.so[.1], libmali.so)
kangaroo_run_opencl: could not load an OpenCL ICD on this host
kangaroo_run_opencl failed (platform 0, device 0) -- no usable OpenCL device?
called boinc_finish(1)

</stderr_txt>

In reply to nekomi_ch's message of 29 Sep 2026:

It mentioned about disk usage limit exceeded, but I have around 40GB of disk space allowed for BOINC, while peak disk usage is just 12.02 MB. Is there any reason as to what happened

<core_client_version>8.2.11</core_client_version>
<![CDATA[
<message>
Disk usage limit exceeded</message>
<stderr_txt>
Keraunos GPU (bounded discrete-log search, CUDA) v1
bitboinc -- https://bitboinc.athena.org.tr/
coded by Alperen Yavuz

</stderr_txt>
]]>

Same problem here

Thanks both for reporting this — found it.

The Keraunos work units' disk-usage bound was set too tight: 10MB, while the GPU build's real peak usage (checkpoint file + distinguished-point table) is closer to 12MB. Once actual usage crossed that estimate, BOINC's client correctly aborted the task with "Disk usage limit exceeded" — not a fault on your end.

Fixed: raised the bound to 200MB for all new work units, and also updated it retroactively on all ~10,000 work units already in the queue, so this shouldn't hit anyone still waiting on pending tasks either. Let us know if it recurs.

---