Since the acceptance tests of the French 5 MV accelerator mass spectrometry facility ASTER in 2007, routine measurement conditions for the long-lived radionuclides 10Be and 26Al have been established. Yearly sample throughput as high as over 3300 unknowns has been reached for 10Be in 2010. Cross-contamination for volatile elements has been largely solved by an ion source upgrade allowing 36Cl measurements at ASTER. However, recent long-term tests using 35Cl/37Cl samples with strongly varying ratios have shown that identical targets lead to different 35Cl/37Cl results at the 2- 4% level when being measured after a time gap of 24 hours while the source is running other samples. Besides time dependent mass fractionation, another very likely reason for this effect might be source memory, thus, asking for sophisticated measurement strategies and improved data evaluation and eventually further ion source improvement. Finally, after establishing quality assurance by cross- calibration of secondary in-house 26Al and 41Ca standards and taking part in round-robin exercises of 10Be and 36Cl, a two-step cross-calibration of secondary in-house 129I standards has been performed. The NIST 3231 standard containing 129I/127I at (0.981 ± 0.012) x 10-6 has been used for step-wise dilution with NaI to produce gram-quantities of lower-level standards for every-day use. The resulting material SM-I-9 (129I/127I: ~1 x 10-9) has been measured vs. AgI produced using minimum chemistry from the two NIST ampoules containing a solution with a nominal ratio 129I/127I of (0.982 ± 0.012) x 10-8. In a second stage, SM-I-10 and SM-I-11 with ratios of ~1 x 10-10 and ~1 x 10-11, respectively, have been cross-calibrated against SM-I-9. Individual uncertainties of the traceable secondary standards are 1.3-1.4 % (2σ), mainly originating from the given uncertainty of the primary NIST 3231 at the 10-8 level. The cross-contamination for iodine is in the range of 0.4-0.6% within the first 20 hours of running the source.